the Massachusetts Institute of Technology and the editors of The Journal of

Interdisciplinary History

Electing Popes: Approval Balloting and Qualified-Majority Rule

Author(s): Josep M. Colomer and Iain McLean
Source: The Journal of Interdisciplinary History, Vol. 29, No. 1 (Summer, 1998), pp. 1-22
Published by: The MIT Press
Stable URL: http://www.jstor.org/stable/205972 .
Accessed: 08/09/2013 14:57

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

The MIT Press and the Massachusetts Institute of Technology and the editors of The Journal of
Interdisciplinary History are collaborating with JSTOR to digitize, preserve and extend access to The Journal
of Interdisciplinary History.

http://www.jstor.org

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
Journal of Interdisciplinary History, xxIx:i (Summer, 1998), 1-22.

Josep M. Colomerand Iain McLean

Electing Popes: Approval Balloting and
Qualified-Majority Rule Social choice theory has ex-
plored the puzzles and paradoxes of voting since 1951. Among its
main findings are that no choice system can satisfy all of the
apparently minimal conditions of fairness and logicality; that every
reasonable electoral system is manipulable; and that when majority
rule fails to produce a unique winner, it may fail catastrophically.1
Some of these points were understood long before the twen-
tieth century. In the eighteenth century, Condorcet was the first
to show that cycles are possible in majority rule. According to
Condorcet, winners are those candidates who prevail over each
of their opponents in exhaustive comparison; yet, such a winner
may not exist. Whenever there are at least three voters and at least
three options, the cyclical majority, A > B > C > A (where >
means "wins a majority against"), is always possible. In the nine-
teenth century, Charles L. Dodgson (aka Lewis Carroll) applied
what we would now call game theory to analyse optimal strategy
for voters and parties. Various writers, culminating in Duncan
Black, realized the interconnection between stability and the
number of issue dimensions. With one issue dimension, and only
mild further assumptions, majority rule is stable. Hence, a good
choice system will select the unique Condorset winner, without
cycling. However, if the number of voting dimensions is two or
Josep M. Colomer is Professor of Political Science at the Higher Council of Scientific
Research, Barcelona, and Georgetown University. He is the author of Game Theoryand the
Transitionto Democracy(Aldershot, I995); PoliticalInstitutionsin Europe(London, 1996). lain
McLean is Professor of Politics, Oxford University. He is the author of "E.J. Nanson, Social
Choice, and Electoral Reform," AustralianJournalof PoliticalScience,XXXI (1996), 369-385;
coauthor, with Alistair McMillan and Burt L. Monroe, of A Mathematical Approachto Propor-
Duncan Black on Lewis Carroll(Boston, 1996);
tionalRepresentation:
McLean would like to thank Haidee Lorrey for invaluable research assistance.
? 1998 by the Massachusetts Institute of Technology and the editors of The Journal of
Interdisciplinary
History.

I Kenneth Arrow, SocialChoiceand IndividualValues(New York, I95I); William H. Riker,

Liberalismagainst Populism(San Francisco, 1982); Allen Gibbard, "Manipulation of Voting
Schemes: A General Result," Econometrica,XLI (1973), 587-601; Mark Satterthwaite, "Strat-
egy-Proofness and Arrow's Conditions," Journal of EconomicTheory, X (1975), 187-217;
Richard D. McKelvey, "General Conditions for Global Intransitivities in Formal Voting
Models," Econometrica,XLVII (I979), IO85-IIII.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
2 JOSEP M. COLOMER AND IAIN McLEAN

more (say, if voters are split by nationality and also, independently,

by religious faction), then simple majority rule is likely to be
unstable and prone to cycling.2
A few ancient and medieval authors-for example, Pliny the
Younger, Ramon Lull, and Nicholas Cusanus-also considered
voting theory in an axiomatic way, and the magistrates of Venice
developed a sophisticated electoral system that was difficult to
manipulate. Their precocious insights into strategic voting and
agenda manipulation-issues that are still current-push back the
invention dates of such procedures as rank-order count, exhaustive
pairwise voting, and approval voting by several centuries. All of
the ancient and medieval choice theorists worked in the context
of practical elections and electoral systems in the Roman Senate
or the medieval Church. This article shows that the Christian
church generated several innovations in voting procedures in the
late Middle Ages, including qualified-majority rule and approval
balloting, and that Church legislators had sophisticated responses
to crucial features of voting. Our sources, where not cited more
explicitly, are the standard collections of medieval canon law.3

FROM DIVINE UNANIMITY TO TWO-THIRDS

MAJORITY During
the late Middle Ages, the Catholic Church devised clear and
efficacious decision rules to protect its autonomy and its temporal
power. The choice of rules to elect the pope was particularly
important, since the Christian church can be viewed as a tradi-
tional monarchy in which the "royal" (papal) powers are based
officially on God's will. For many centuries, lack of rules provoked
2 M. J. A. Nicholas de Caritat, Marquis de Condorcet, Essai sur l'application de l'analysea la
probabilitedes decisionsrenduesa la pluralitedes voix (Paris, 1785); McLean and Fiona Hewitt,
Condorcet:Foundationsof Social Choiceand PoliticalTheory (Brookfield, Vt., I994); McLean,
AlistairMcMillan, and Burt L. Monroe, A Mathematical Approachto ProportionalRepresentation:
Duncan Black on Lewis Carroll (Boston, 1996); Duncan Black and Ronald A. Newing,
CommitteeDecisions with ComplementaryValuation(Glasgow, I951); Black, The Theory of
Committeesand Elections(Cambridge 1958).
3 McLean and John London, "The Borda and the Condorcet Principles: Three Medieval
Applications," Social Choiceand Welfare,VII (I990), 99-I08; McLean and Arnold B. Urken,
Classics of Social Choice (Ann Arbor, 1995); Marji Lines, "Approval Voting and Strategy
Analysis:A Venetian Example," TheoryandDecision,XX (I986), 155-172. For medieval canon
law, see also Angelo M. Cherubini (ed.), BullariumRomanumnovissimum(Rome, I638); Emil
Friedberg, CorpusJuris Canonici(Leipzig, 1879); Paul Fabre and Louis Duchesne (eds.), Le
Liber Censuum de l'Eglise romaine(Paris I889-1952), 3v.; Norman Tanner, Decreesof the
EcumenicalCouncils (Washington, D.C., 1990), translated from ConciliorumOecumenicorum
Decreta(ed. Giuseppe Alberigo et al.) (Bologna, I973; 3d ed.).

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
ELECTING POPES | 3

frequent conflicts and schisms, leaving the Christian church under

the protection and domination of political powers, especially
emperors and Italian noble families. Selecting rules to elect the
pope was the result of a several-century process of clarifying (I)
who the voters were; (2) how the ballots were to be organized;
(3) what decision rules were to be used; and (4) what incentives
the procedure was to provide for a sound and quick decision,
avoiding strategic maneuvers or long pontifical vacancies.

EMPEROR VERSUS POPE Peter, the first "pope" of the Christian

church, was chosen by the founder of the religion, but apparently
Jesus did not prescribe the rules for appointing Peter's successors.
Since the pope was the bishop of Rome, whose primacy over the
other bishops was only gradually asserted, the procedure of his
election was initially similar to that of the other bishops. In certain
local churches, it was customary for bishops to choose their
successors. Although this practice was forbidden by the council
of Antioch in 341 and again by the council of Rome in 465, it
survived in Rome until the sixth century. Thereafter, popes oc-
casionally expressed opinions about their preferred successor, but
bishops usually were elected by the Christian faithful ("the peo-
ple')-the clergy and the bishops of the province-under the
chairmanship of the metropolitan.
The election of a bishop was conceived as a way to discover
God's will. It was guided by unanimity rule, the only rule that
could assure the participants that their decision was right. Hence,
the maxim "vox populi, vox Dei" (the people's voice is God's
voice). The people could propose candidates, but their choices
had to be accepted by the bishops of the province and the
metropolitan. Thus, popular participation induced consensus and
obedience from the faithful rather than opening up the set of
choices.
The Pope was elected according to this system for a while.
First, lay members of the Roman Church-especially nobles,
military officers, magistrates, and other high officers of the
Empire-offered several candidates viva voce. Then, the clergy
proceeded to reduce or to enlarge that list, and, finally, the sixteen
bishops of the Roman province settled the matter. This sequence
was reflected in the motto of Pope Leo I (440-461): "vota civium,
testimonia populorum, honoratum arbitrium, electio clericorum."

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
4 JOSEP M. COLOMER AND IAIN McLEAN

This process frequently produced conflicts and schisms. Be-

fore Christianity had been officially accepted by the emperor, at
least one simultaneous election of two different popes had taken
place (in 250, after eighteen months of failure to elect). In 366
and 418, the election of two popes by different factions of the
Church provoked hundreds of deaths and the intervention of
Roman troops. These conflicts put the Church under political
protection for several centuries. Honorius, the emperor, ruled in
420 that if two popes were elected, neither would be valid, and
a new election would be held and unanimity required.
A violent, bloody schism in 498/499 induced one of the new
popes, Simaccus, to decree that, in the absence of a nominated
successor from a deceased pope, a new pope should have the
unanimous support of the clergy or, in case of division, the support
of the majority. This relatively early attempt to exclude the lay
powers and to institute a non-unanimity decision rule could not
avert the conflict between a nominated successor and an elected
candidate soon thereafter, resulting in the reinstatement of the
emperor as arbiter.
Although most of the popes from the fourth to the twelfth
century were appointed or confirmed by the emperor, imperial
power had begun to dwindle by the eleventh century. "The
election decree of I059 gave the cardinals of the Roman church
the decisive part in papal elections, and reduced the emperor's
role to a shadow." A decisive shift came with the submission of
Emperor Henry III to Pope Gregory VII at Canossa in 1077,
followed by the emperor's formal renunciation of the privilege at
the Concordat of Worms in II22.4
During the period of imperial control, the quality of those
appointed declined, but the emperor always managed to obtain
enough support for his candidates to make unanimous agreement
on an alternative impossible. Twice in the ninth century, when
the former imperial power had weakened, the Church obtained
an emperor's renunciation of the right to appoint the pope. New
conflicts and schisms, however, immediately led to the reimposi-
tion of imperial control.

4 H. E. J. Cowdrey, The EpistolaeVagantesof GregoryVII (Oxford, I972), xvii.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
ELECTING POPES | 5

The present Catholic church recognizes 159 popes from I to

1122 A.D., but at least 31 "anti-popes" were recognized by certain
factions. From the mid-ninth to the mid-tenth century, from a
total number of twenty-six popes, twelve were removed from
office, five sent into exile, and five killed.5

THE DIFFICULT UNANIMITY OF DIVINE RIGHT Pope Gregory

VII's Dictatus Papae (1075) asserted the primacy of the pope over
political powers. Sometimes called a bull, Dictatus Papaewould be
better described as a set of headings for a projected collection of
canons. Enforcing papal primacy required intensive legislative
activity. It started with Gratian (II39-II40), and the following
Ioo years saw papal approval of four enlarged canon law codes.
Gregory's correspondence illustrates "the practical application of
the maxims of papal authority, which Gregory had laid down in
the Dictatus Papae," although scholars have pointed out that many
of the later compilations did not cite Gregory.6
After 1311, no new canon law codes were adopted until 1918,
and again in 1983. The canonists wrote extensively about election
procedures, but most of them do not focus on voting rules. Nor
do they seem to contain many significant innovations in practical
terms. The Church required an orderly succession of its monarch.
New rules emerged from successive decisions about partial aspects
of the question. The first was a papal bull of Io59-also a canon
of the Council of Rome-that excluded the laity from the elec-
tion of the pope (Nicholas II, In nomine Domini). It reserved
priority in the election to the cardinal-bishops, who were en-
trusted with gaining the assent of the cardinal-priests and the
cardinal-deacons, as well as the approval of the other members of
the clergy and the people. The role of the emperor was reduced
once again to mere acceptance of the Church's decision. Although
France, Germany, and Spain tried to exercise their veto against
5 All of these numbers remain the subject of some dispute. See Gaetano Moroni, "Cardinal,"
in Dizionariode erudizionestoricoecclesiastica
(Venice, I89I), IX, 272-318; X, 5-26; T. Ortolan,
"Election des Papes," in Alfred Vacant and E. Mangenot (eds.), Dictionnairede theologie
catholique(Paris, 1939), IV, 2282-2319. Specific data for the calculations in this study are in
Jean Gaudemet, Eglise et cite. Histoiredu droitcanonique(Paris, 1994).
6 Cowdrey, EpistolaeVagantes,xxvii. Cf. J. Gilchrist, "The Reception of Pope Gregory VII
into the Canon Law (1073-II4I)," Zeitschriftder Savigny-Stiftung kanonis-
fur Rechtsgeschichte,
tischeAbteilung,LIX (I973), 35-82.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
6 | JOSEP M. COLOMER AND IAIN McLEAN

certain candidates for pope, this tactic now had to be implemented

through some faction of cardinals.7
This procedure did not avert conflict either. Three twelfth-
century elections produced eight anti-popes and only nine
"official" popes in less than fifty years. In the absence of direct
appointments, the Church came up with three ways to achieve
unanimous agreement in the event of dissension among the elec-
torate: "acclamation," "scrutiny," and "compromissum." In I215,
the Fourth Lateran Council instituted them for all Church elec-
tions, in its document Quia propter.
Acclamation, also known as "quasi-inspiration," corre-
sponded most closely to the expectation that voters would have
divine inspiration. Yet, elections by acclamation were rare, and
even participants complained that they were induced more by the
enthusiastic and threatening roar of the crowds than by any natural
rallying of voters around a single candidate. Compromissum sig-
nified the delegation of final choice to a small commission when-
ever long sessions and repeated failures showed unanimous
agreement to be unlikely. However, since this option had to be
adopted by unanimity as well, it was not a frequent resource
either.8
Scrutiny, that is, voting, was the most common procedure,
but new intellectual devices were needed to create apparent una-
nimity when it did not exist. The most noteworthy was the sanior
et maior pars-the "sounder and greater part." According to a

7 For canonists' writing on voting rules, see especially Gerardus Franses (ed.), Summa
'Elegantiusin iure divino"seu Coloniensis(1169) (New York, I969); Ernest Adolph Theodor
Laspeyres (ed.), BernardiPapiensis,SummaDecretalium,cap. II, 'BernardiSumma de Electione"
(1178)(Graz, 1956); Romualdo Trifone (ed.), "Guiglelmo Nassone, Lecturaee Summae,section
'Ad quod capitulum vacante sede' (c. 1234)," Rivista di Storiadel Diritto Italiano, II, (I929),
242-260; Paul Viollet (ed.), "Guillaume de Mandagout, Tractatusde electionibus (c. I285)," in
Histoirelitterairede la France(Paris, 1914), xxxiv; Henrici de Segusio, Cardinalis Hostiensis,
SummaAurea, "De electione, & electi potestate" (I574-158I) (Turin, 1963; new ed.); Primum
DecretaliumLibrumCommentaria,"Electione, & electi potestate" (158I) (Turin, 1965; new ed.).
We have also examined writings by Ruffinus (1157-1159), Stephanus Tornacensus (c. I160),
BernardusCompostellanus (c. 1178), Simon von Bisignano (c. 1179), Panormitanus(Nicholas
de Tudeschis), and Baldo degli Ubaldi (c. I350). Johannes Dominicus Mansi (ed.), Sacrorum
conciliorum nova et amplissimacollectio(Venice, 1759-1793), XIX, 897, 907.
8 For the election of Gregory VII in I073, as reported by the pope himself, see Jacques-Paul
Migne (ed.), PatrologiaLatina (Paris, I853), CXLVIII, 643-734; Alexander Murray, "Pope
Gregory VII and His Letters," Traditio,XXII (1966), 149-201; Ian S. Robinson, The Papacy
1073-1198: Continuityand Innovation(Cambridge, 1990), 59-60.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
ELECTING POPES | 7

(rather reiterative) modern literature about canon laws and exe-

geses, "saniority" referred not only to the priority role given to
cardinal-bishops, but also to the merits of candidates and voters-
their zeal and dignity (including age and length of time in official
capacity, that is, seniority and hierarchy). These qualities were
considered relevant because voting was conceived as a way to
discover the truth. But the "sanior pars" often did not coincide
with the "maior pars."In lower elections, the determination about
which was the "sanior" part in a divided electorate was usually
submitted to an arbiter, such as the metropolitan bishop or even
the pope, but no such arbiter existed for the election of the pope
after II22.9

A TYPICAL SCHISM The election of I 130 illustrates the difficul-

ties of applying the unanimity rule. As shown in Table I, thirty-
eight cardinals participated. Gregoria Papareschi, one candidate,
was a cardinal-deacon and member of the Frangipani family,
which had been protected and promoted by Honorius II, the
previous pope. Pietro Perleoni, his rival, who was popular in
Rome, was a cardinal-priest and also a member of a noble fam-
ily-this one favored by Calixtus II, an earlier pope.
Initially, the election was made in compromissum by a com-
mission of eight cardinals. The college agreed to recognize the
commission's choice, and each of the three orders of cardinals in
the college elected its own representatives. The commission de-
cided in favor of Papareschi, afterward called Innocent II, by five
votes to three.
The members of the commission tended to prefer their own
kind. Most of the cardinal-deacons and cardinal-bishops sup-
ported Papareschi, and most of the cardinal-priests supported

9 Adhemer Esmein, "L'unanimiteet al majorite dans les elections canoniques," in Melanges

Fitting (Montpellier, 1907); Edoardo Ruffini Avondo, "I1principio maggioritario nella storia
del Diritto Canonico," ArchivioGiuridico'FilippoSerafini,"IX (1925) 15-67; Leo Moulin, "Les
origines r6ligieuses des techniques electorales et deliberatives modernes," Revue Internationale
d'HistoirePolitiqueet Constitutionelle,III (I953), o06-I48; idem, "Sanior et maior pars. Studio
sull'evoluzione delle tecniche elettorali negli ordini religiosi dal VI al XIII secolo," Studi
Politici,VI (I959), 48-75; Alexis Petrani, "Genese de la majorite qualifi6e,"Apollinaris,XXX
(I957), 430-438; Gaudemet, "Unanimite et majorite (Observations sur quelques etudes
recentes)," in Etudes historiquesa la memoirede Noel Didier (Paris I960), 149-162; Robinson,
Papacy.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
8 | JOSEP M. COLOMER AND IAIN McLEAN

Perleoni-both in the whole college and in the commission-in

proportions of three to two. However, as Table I shows, even
though a majority of the commission-its members elected sepa-
rately by the three orders of the college-preferred Papareschi, a
majority of the college preferred Perleoni. This discrepancy led a
majority of cardinals to reject the election by the commission and
to elect their preferred candidate instead, now called Anacletus II.
The existing rules were not much help. Although Papareschi
(Innocent II) enjoyed the favor of the majority of the sanior
pars in the sense of higher dignity-that is to say, the cardinal-
bishops-Perleoni (Anacletus II) had the support of the maior
pars, including the senior (oldest) members of the college.
Innocent had to flee Rome, but he commanded political
support in France, Germany, and England, and Church support
from bishops in Spain and Lombardy. Anacletus was recognized
in Rome, Sicily, and Scotland, but was considered the anti-pope.
Thus, as during other disputes, some monasteries had two abbots
and two priors, some parishes two priests, and so on. The schism
lasted for eight years.10

Table1 Divisions in the PapalElection of II30

VOTES BY CANDIDATE, VOTES BY CANDIDATE,

COLLEGE COMMISSION

PAPARESCHI PERLEONI PAPARESCHI PERLEONI

VOTERS (DEACON) (PRIEST) (DEACON) (PRIEST)
Cardinal-bishops 6 4 2 2 o
Cardinal-priests 23 7 16 I 2
Cardinal-deacons 9 6 3 2 I

Total 38 17 2I 5 3
Winner Perleoni (Anacletus II) Papareschi (Innocent II)
NOTE Exact data of voters for each candidate in every order of cardinals-including the
names of voters and the pope who had nominated them-can be found in Onuphrio
Panvinio, EpitomePontificorumRomanorum. . . (Venice, 1557): 99-Io6; and "Romanorum
Chronicum," in HistoriaB-Platinaede VitisPontificorum(1611). Modern authorsare less reliable.

Io Our source for this account is Onuphrio Panvinio, EpitomePontificumRomanorum.A S.

Petrousquead PaulumIV. Gestorum(videlicet) Singulorum,& ConclaviumCompendiaria
electionisque
Narratio(Venice, I557). See also Franz Josef Schmale, Studien zum SchismadesJahres 1130
(K6ln, I96I).

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
ELECTING POPES 9

THE EFFICACIOUS BUT SLOW RULE OF TWO-THIRDS Such

conflicts led to the adoption of two-thirds majority rule by Pope
Alexander III (himself previously in competition with an anti-
pope) during the Third Lateran Council in 1179. The rule of
two-thirds had once been operative in certain Roman municipal
curia albeit not as a decision-making instrument but as a quorum
requirement-as well as in the election of abbots. This and other
qualified-majority rules also saw use in several Italian communes
of the Middle Ages. Alexander III's six-month stay in Venice
during I 77, forging reconciliation with the Emperor Frederick
Barbarossa who had supported the anti-pope, may well have
exposed him to the city's sophisticated voting procedure to elect
a doge and induced him to adopt this more efficient, non-
unanimity rule for elections in the Church soon thereafter.11
Together with the new decision rule, the canon of 1179
stipulated equal voting rights for the members of the three orders
of cardinals (abolishing the priority of the cardinal-bishops). The
division of voters into three parts, from which the agreement of
two was required, was not arbitrary.The scrutineers and inspectors
of election returns had to comprise one member from each of the
three orders. Giving equality to all of the cardinalsin voting might
have been an indirect way to suggest that a coalition of members
from at least two of the three orders would be necessary. This
interpretation is also a possible reading of the text, which states
that the pope will be the candidate elected and accepted by "two
parties" (Licetde evitanda,1, io). Yet, the basic aim of the qualified
majority of two-thirds seems to have been the formation of a large
coalition of cardinals. Hence, a two-thirds winner often would
have resulted from a compromise between supporters of different
candidates. As contemporary analyses note, once a candidate was
elected, the losers would have had to persuade a majorityof the
winner's original supporters to change their mind. Faced with this
requirement, the losing coalition could hardly have been expected
to continue the fight.12

1 Ruffini Avondo, "I1principio maggioritario";idem, I sistemidi deliberazionecollettivanel

medioevoitaliano(Turin 1927); LadislasKonopczynski, Le Liberumveto.Etudesurle developpement
(Paris, 1930); MarshallW. Baldwin, AlexanderIII and the TwelfthCentury
duprincipemajoritaire
(New York, 1968); Murray, Reason and Society(Oxford, I978); Lines, "ApprovalVoting."
I2 Alberigo et al. (eds.) OecumenicaConcilia,2; Donald G. Saari, Geometryof Voting(Berlin,
1994), I5-I6.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
10 JOSEP M. COLOMER AND IAIN McLEAN

Recent research in social choice theory has proven the in-

vulnerability of the two-thirds rule to disequilibrium or cycles.
Specifically, Caplin and Nalebuff have shown that the rule of 64
percent guarantees a single winner under conditions of concavity
in voter preferences-meaning that, when more voters prefer
intermediate candidates than the average of those favoring ex-
tremes, an unbeatable proposal exists, and no cycles are possible.
In general, the majority rule needed to ensure such a situation in
a n-dimensional issue space is no higher than i-[n/(n+I)]n. Hence,
this rule is equal to 55 percent for two-dimensional spaces and to
57 percent for three-dimensional spaces, and, as n increases, its
limit is just under 64 percent.
To approach the matter differently, for three candidates, "it
is impossible to have a cycle where each candidate beats another
candidate by receiving more than 2/3 of the vote." The general
rule that makes cycles impossible is equal to (n-)/n---n, in this
case, being the number of candidates. Hence, the two-thirds rule
used to elect the pope produces a stable outcome with up to three
candidates, a number that might be expected from a college of
voters with three orders. Although electors did not necessarily
vote by order in blocs, three just happens to be the largest number
of simultaneous popes to have been appointed. Saari has derived
the following condition for generic stability: q-majority rule is
generically stable iff k < zq - n, where q = the threshold number
required to carry a proposal; k = the number of issue dimensions;
and n = the number of votes cast.13
Note that the formal findings in social choice theory just
presented refer to the number of dimensions in the issue space
and to the number of candidates. However, both dimensions and
candidates are, to some extent, endogenous to the decision rule.
If doctrinal or ideological allegiances are not restrictive for voters'
choices (as in many papal elections), cardinals likely to lose have
an incentive to introduce new candidates defined by personal
character or other idiosyncratic dimension, in order to divide

13 Issue spaceis a technical term from social choice theory. If opinions are divided along
more than one dimension-for instance, about religious belief and about nationality-they
can be mapped onto a diagramwhere one axis (say, the x axis) representsthe possible positions
on issue I and another (say, the y axis) represents the possible positions on issue 2, thus
creating an issue space. The positions of individuals within that space can be mapped using
Cartesian coordinates.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
ELECTING POPES | II

voters' preferences further and force larger and more inclusive

bargains.14
Thus did mathematical precision replace previous discussions
about subjectively estimated qualities of candidates and voters. As
Pope Gregory X stated, "Non zeli ad zelum, nec meriti ad
meritum, sed solum numeri ad numerum fiat collatio" ("Not zeal
to zeal, nor merit to merit, but solely numbers to numbers are to
be compared." VI? Decretalium,lib. I, tit. VI, cap. 9). Popes and
canonists identified the maior pars with the sanior pars. In the
words of Pope Pius II about his own election in 1458, "What is
done by two thirds of the sacred college [of cardinals], that is
surely of the Holy Ghost, which may not be resisted."15
This quantitative approach was also consistent with other
legal provisions of the time: first, with secret vote (finally adopted
as the formal rule in the sixteenth century, although used earlier),
which makes any discussion about the qualities of voters impos-
sible; and second, with the rule against absentee votes, which
resulted not so much from expediency as from the idea that people
were more apt to make informed decisions when they are to-
gether, which prefigured the arguments of Rousseau and Con-
dorcet about majority choice. The two-thirds rule was later
adopted for other procedures in the election of the pope. In an
election by compromissum, although unanimity was the standard,
two-thirds of the commission was sufficient to decide. In accla-
mation, at least two-thirds of the cardinals had to participate in
any tumultuous, decisive demonstration for a candidate. The rule
of two-thirds also came into play in elections of bishops and
abbots.'6
The two-thirds rule produced the desired stability, but it had
unintended consequences. Negotiations to form a sufficiently
large coalition provoked extremely long delays and vacancies. The

I4 Andrew Caplin and Barry Nalebuff, "On 64%-MajorityRule," Econometrica, LVI (1988),
787-814; idem, "Aggregationand Social Choice: A Mean Voter Theorem," ibid., LIX (I991),
I-23; Saari, Geometryof Voting,92-93; idem, Basic Geometryof Voting(Berlin, I995), 62; idem,
"The Generic Existence of a Core for q-rules,"Journalof EconomicTheory,IX (1997), 219-260;
Riker, The Art of PoliticalManipulation(New Haven, 1986); idem(ed.), AgendaFormation(Ann
Arbor, 1993).
I5 Florence A. Gragg and Leona C. Gabel (eds.), Memoirsof a RenaissancePope: The
Commentaries of Pius I (New York, 1959), 88.
I6 Jean-Jacques Rousseau, Du ContratSocial (Paris, I762), II, iii and IV, ii; McLean and
Hewitt, Condorcet,34-43.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
12 [ JOSEP M. COLOMER AND IAIN McLEAN

electors in I216, I24I, I243, I26I, 1265, and 1268-1270 took

several months to reach a decision, in several cases resorting to
compromises in commission. In 1243 (and, according to Ruffini
Avondo, in 1216) the civil authorities, under popular pressure,
reacted to cardinals' slowness by locking them up. In I241, the
senator of Rome (that is, the head of civil administration) confined
the cardinals to an old, unhygienic building, guarded by police.
He was able to elicit a decision only by threatening to have the
corpse of the dead pope exhumed and shown publicly in full
regalia. After another deliberation that started in 1268 and lasted
more than two years without an agreement, the public besieged
the cardinals in the episcopal palace, removing the roof and
allowing nothing but bread and water to go inside. According to
a contemporary joke, the roof was removed in order to let "the
Holy Spirit receive an unhampered entrance." A new pope was
elected on this occasion by compromissum after a record vacancy
of 34 months. The two-thirds rule produced efficacious and rather
stable outcomes, at the price of long delays in decision making.
This is now recognized as a classic trade-off in social choice.17

THE SURPRISING CONCLAVE The practice of locking cardinals

up led Pope Gregory X to establish a formal procedure for their
seclusion, known as the conclave (from the Latin, with key); it was
approved by the council of Lyon in I274, under the name, Ubi
periculum.Since Gregory had not been a cardinal before his elec-
tion, "he did not take the college's interests much into considera-
tion. " Ubi periculumaimed to obtain a quick decision and prevent
strategic maneuvering in papal elections. Similar institutions had
been established in the Dominican constitution of 1228, as well
as in such communes as Venice and Piacenza in 1229 and 1233,
respectively.18
The conclave had the cardinals gathered in a single room in
the papal palace, the doors of which were walled up and secured
by soldiers. No cardinal was to have more than two servants with
17 Ruffini Avondo, "I1 principio maggioritario." See also Peter Herde, "Election and
Abdication of the Pope: Practice and Doctrine in the Thirteenth Century," Proceedings
of the
CongressofMedievalCanonLaw,VII (1985), 41 I-436; Andre Vauchez, Apogee
Sixth International
de la Papauteet expansionde la Chretiente(1054-1274)(Paris, 1990).
I8 Herde, "Election and Abdication," 418; Ruffini Avondo, "I principio maggioritario";
Walter Ullmann, A ShortHistoryof the Papacyin the MiddleAges (London, 1972).

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
ELECTING POPES | 13

him or any communication with the outside world. Food entered

through a guarded window. After the fourth day, the menu was
restricted; after the ninth, it was reduced to bread, water, and
wine. The cardinals received no income until they reached a
collective decision.
Although some of these hardships were later softened, they
motivated the cardinals to reach a common decision. Whether
lack of nutrition was a factor is uncertain, but the illness, and
sometimes even the death of cardinals (six cardinals died during
the election of 1287 alone), tended to precipitate agreement
among the remaining participants. Since bribery, coercion, or
explicit arrangement between cardinalswas forbidden under pen-
alty of excommunication and annulment of the election, large
coalitions were difficult to build, and candidates were often
elected on the basis of their immediate, apparent appeal rather
than careful evaluation of their merits or religious characteristics.
The first election of a pope under this procedure, in 1276,
took only one day. The following popes suspended the application
of conclave, whereupon long delays reappeared: more than seven
months in I277, six months in I28I, almost eleven months in
1288, and twenty-seven months from 1292 to 1294. This counter-
proof of conclave's efficacy moved Celestine V, the pope elected
in I294, to reestablish it.

THE UTILITY FUNCTION OF CELESTINE V The twenty-seven-

month deadlock of 1292 to I294 was broken when the name of
Peter of Morrone, the "octogenarian hermit," was introduced. His
election seems to have resulted from a snowballing effect-more
and more electors clutching at a candidate, however unlikely, to
break the impasse. Although the official reason given for his
election was "inspiration," we cannot reconstruct how much
admiration of his sanctity, desire to end the stalemate, or calcula-
tion that any error would be short-lived, entered into it.19
The hermit pope took the name, Celestine V. His term of
office was short but momentous. Elected in July I294, in Sep-
tember he reintroduced the conclave regulations of Gregory X;
in December, he abdicated and returned to his hermit's cell.
Celestine V was even more of an outsider than Gregory X. Not

I9 Herde, "Election and Abdication," 427.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
14 JOSEP M. COLOMER AND IAIN McLEAN

only had he not been a cardinal before becoming the pope; his
behavior afterward suggests that he derived no particular benefit
from his position. A pope who had previously been a cardinal,
and/or needed the cooperation of the cardinals for his continued
comfort in office, might not have wanted, or been able, to impose
the reforms of Gregory X and Celestine V. But Celestine's unusual
provenance afforded him a certain immunity. As with other con-
stitutional actors (such as Thomas Jefferson and James Madison),
Celestine was able to write a successful constitution because he
had little personal interest in exploiting his own office. Short,
successful conclaves lasting a few days or less have become the
norm since Celestine's election.20
The effects of the conclave on the winning candidates were
puzzling. Frequent surprises gave rise to the popular saying, "He
who enters the conclave a pope, leaves it a cardinal," which
achieved the status of an "empiricallaw" in the fourteenth century.
In the "great schism" of Avignon in 1378, the same cardinalswho
had elected the pope in conclave under strong political pressures
cancelled their decision on grounds of duress once they realized
the kind of man that they had chosen. Within a few months, they
chose a new one.
According to the basic rules of Gregory X and Celestine V,
which are still enforced, (I) voters must be cardinals; (z) the
decision threshold must be set above two-thirds of the voters; and
(3) the cardinals must be secluded in conclave in order to reach
a quick and sound decision. These policies embodied successive
partial reforms, each adopted in reaction against unintended effects
of previous decisions. The Church's legislators and the canonists
seem to have refrained from thinking about voting too technically
out of the belief that outcomes ought to correspond to God's will.
Yet, they had to face real problems in voting matters by importing
or inventing ad hoc devices for particular aims.

APPROVAL BALLOTTING IN THE FOURTEENTH CENTURY From the

late thirteenth to the early seventeenth century, the ballot used in

20 The rules of conclave, their enforcement, and successive modifications can be found in
DJ.C.B., Los Papas y el C6nclave(Barcelona, 1878); Lucio Lector, Le conclave(Paris, 1894);
Ortolan, "Conclave," in Vacant and Mangenot (eds.), Dictionnairede theologiecatholique(Paris,
1938), III, 707-727. Other rules about the timing of the election are in OrdoRomanus,XII.
See Migne (ed.), PatrologiaLatina (I844/I845), LXXVIII, cols. 1063-II06, esp. I097-IIOO.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
ELECTING POPES 15

papal elections by the cardinals was a form of what is now called

"approvalvoting." This rule is not contained in any papal bull or
decree, but in a lower-rank set of provisions-the religious
equivalent of parliamentary rules of order-called Ordinarium
SanctaeRomanae Ecclesiae,written by Jacobi Gaytani a short time
after the definitive enforcement of the conclave rules. Gaytani, a
protege of Pope Boniface VIII, who nominated him as a cardinal,
was the cardinal-deacon in charge of the Church of Saint George
of the Golden Veil in Velabro. He attended five conclaves from
1305 to 1352.21
Gaytani specified the voting procedure, both by scrutiny and
by compromissum, the latter acceptable only in case of need. In
voting by scrutiny, the ballot is so written that the voter can
choose either one or several candidates. Gaytani advised the car-
dinals not to choose many candidates "for decency and expedi-
ency," although there is no formal restriction about the matter.
Election results take into account all nominations received by
every candidate. This form of ballot seems to have been used in
the forty-one conclaves held from 1294 to 1621 (the last conclave
before new regulations establishing the categoric vote-Eterni
Pacis, 1621, and Decet Romanum Pontificem,1622).
According to social choice theory, approval voting (Av) tends
to promote consensual, relatively high social-utility winners; it
satisfies several relevant criteria, such as monotonicity; and it is
relatively easy to implement. Although Av has some pitfalls-for
example, the inability to guarantee that a Condorcet winner or a
majority winner will win-its advantages make it a highly desir-
able procedure for single-winner elections with more than two
candidates-as in the election of popes.

APPROVAL BALLOTING WITH TWO-THIRDS MAJORITY RULE Ap-

proval voting is usually associated with single-winner elections,
the winner being the candidate with the most votes. Using it with
two-thirds majority rule can have bizarre consequences; it can
2I Steven Brams and Peter Fishburn, "Approval Voting," AmericanPoliticalScienceReview,
LXXII (1978), 831-847; idem, Approval Voting (Boston, 1983); Hannu Nurmi, Comparing
VotingSystems(Dordrecht, 1987); Samuel Merrill, MakingMulticandidate ElectionsMoreDemo-
cratic(Princeton, 1988). Gaytani'swritings about procedures in papal elections can be found
in Johanne Mabillon and Michaele Germain, Musei Italici. Tomus II. ComplectensAntiquos
LibrosRituales SanctaeRomanaeEcclesiae(Paris, 1689), 245ff. See also Moroni, "Cardinals."
who claims that Gaytani was Boniface's nephew.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
I6 JOSEP M. COLOMER AND IAIN McLEAN

give several candidates more than the required proportion of votes

to win. This undesirable occurrence, which nullifies an election,
is more probable the smaller the number of candidates, the more
homogeneous the preferences of voters regarding candidates, and
the greater the incidence of voting for multiple candidates. In an
extreme case, each of two candidates can obtain two-thirds of the
approval votes when only one-third of voters choose more than
one candidate, if each of the two candidates that these voters
approve is the single preference of another third of the voters.
Approval voting can also produce no winner. The probability that
no candidate will obtain two-thirds approval is greater the higher
the number of candidates, the more dispersed the preferences of
voters, and the smaller the number of approval votes on the ballot.
In conclaves that failed to elect a candidate, further rounds of
approval voting were implemented. There was no elimination
threshold between one round and the next; candidates were
always eligible even if they had not obtained a single vote in
previous rounds.
A second relevant device is the vote in accessit,or "access,"
which made its earliest appearance in the I455 conclave. If no
candidate obtains a sufficient majority, a voter can add a candidate
or candidates for whom he has not previously voted but who has
received votes in the first ballot. Precedent for this tactic might
have been a practice of the Roman senate, whereby a senator who
wanted to share a position with another physically moved to join
the proposal's promoter, giving an immediate visual display of
majority and minority support in any binary option. The cardinals
began access rounds with public declarations of support, followed
by a scrutineers' certificate that the new proponents had not
previously voted for the same candidates. To facilitate proof, every
ballot was marked with a personal but anonymous stamp and
included the voter's signature folded inside. To preserve voting
secrecy even from scrutineers, however, the access round was soon
transformed into a new voting round with written ballots on
which the cardinals could choose between giving new approval
to one or several candidates and confirming their previous
choice.22

22 Betty Radice, The Lettersof the YoungerPliny (Harmondsworth, 1969), 220-224; Riker,
PoliticalManipulation,78-88.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
ELECTING POPES 17

The procedure in access rounds was similar to that in the

preceding round with approval ballot, except that only candidates
who had already obtained at least one vote were eligible for
additional approval. Every access round was tied to a voting
round, and every failure to determine a winner produced succes-
sive rounds of voting and access that provided voters with dy-
namic information about others' preferences and their collective
results.
Access counterweighs some of the expected consequences of
approval balloting, especially by altering voters' strategies to select
a winner, but also by prolonging the number of voting rounds.
In order to model voters' rational strategies under approval bal-
loting, we impute trichotomous preferences to every voter-(I)
preferred candidates, (2) acceptable (though not preferred) candi-
dates, and (3) nonacceptable candidates-allowing that some of
these categories may be void for some voters. Approval voting
with the usual plurality rule permits the identification of two
strategies: (SI) Voters who prefer candidates with slim or uncertain
chances to win always vote their preferred candidates and usually
vote their acceptable ones. (S2) Voters who prefer candidates with
good chances to win vote only for their preferred candidates.
As a result of these strategies, winners receive all of the
support of those who prefer them and most of the support of
those who find them acceptable. Candidates with extreme views
win only if their majority followers feel free to vote for them
because none of their nonacceptable candidates has a chance to
win, that is, they enjoy the support of a large, compact-even
biased-majority facing several dispersed minorities. However,
under approval balloting with access, strategic voters should ap-
prove fewer candidates. A voter who can express approval of an
acceptable candidate in a later round-as in access-can hold back
in the first round to determine whether his preferred candidate is
likely to win. Voters tend to use strategy S2, however, increasing
the probability that no candidate will obtain sufficient votes to
win.
Given scanty information regarding other voters' preferences
and candidates likely to win, as were the cardinals when they
arrived in conclave, early results resemble polls in mass elections.
Voters whose preferred candidates appear not to have sufficient
backing in early rounds can expand their set of approved names-

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
I8 JOSEP M. COLOMER AND IAIN McLEAN

including the most acceptable, or least objectionable, among the

front-runners. Yet, as Brams and Merrill have noted, poll infor-
mation also gives voters incentives to "truncate" sincere prefer-
ences, that is, to vote for fewer candidates than those they find
acceptable, thus hastening the outcomes.23

ECCLESIASTICAL ORDERS AND POLITICAL POWERS The difficulty

of reaching agreement in most papal elections was accentuated by
the numerous candidates and the cardinals'dispersed preferences.
The greater the number of parties and candidates, the more likely
it was that an unexpected candidate would become pope. In late
medieval and early modem times, ecclesiastic orders (bishops,
priests, and deacons) and political allegiances were the variables
that complicated papal elections.
Since the end of the eleventh century, the college of cardinals
has comprised seven cardinal-bishops, twenty-eight cardinal-
priests, and eighteen cardinal-deacons-a total of fifty-three and
a two-thirds majority threshold of thirty-six. Hence, in spite of
its supreme rank in the hierarchy, the order of cardinal-bishops
had no electoral power by itself, even if it acted as a unit. The
members of the two other orders could form a more than
sufficient majority, whereas a coalition of bishops and either of
the two other orders was not sufficient to win. This situation
probably moved Pope Nicholas II to confer "principaljudgment"
to the cardinal-bishops (as noted above). Yet, further levelling,
when the rule of two-thirds was established, apparently motivated
the cardinal-bishops to position themselves as leaders of different
factions of cardinal-priests and cardinal-deacons. In I585, the
members of the college increased to seventy, but the new com-
position (six bishops, fifty priests, and fourteen deacons) main-
tained the same incentives to form factions.24
Faction leaders bartered their support for future appoint-
ments, castles, palaces, wealth, or income whenever possible. In
some cases, they were even able to maneuver elected candidates
into honoring platforms that severely limited their powers-
known as "capitulations." Three or four faction leaders usually
23 Elections;Brams, "Polls and the Problem of Strategic Information
Merrill, Multicandidate
and Voter Behavior," Society,XIX (1982), 4-II.
24 Data in this paragraphare from Ullmann, Historyof the Papacy,232; Moroni, "Cardinals,"
287.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
ELECTING POPES I19

appeared per conclave, although new coalitions and splits from

one conclave to another were normal.25
The second basic source of division among the cardinals was
their different political allegiances with the main Christian king-
doms in Europe, especially France, the German confederation,
and Spain. The council of 1417 in Constance put an end to the
"great schism" of Avignon. The twenty-three cardinals of the
college were joined by six other bishops for each of the five
nation-groups (French, German, English, Spanish, and Italian),
creating the total of fifty-three members in the decision body, in
order to obtain a wide consensus. It also established the two-thirds
rule for the college of cardinals and every group of "national"
deputies. The ambassadors from the great powers often visited
conclaves to transmit national opinions about candidates.26
The emergence of six or seven factions-occasionally as many
as ten-at the beginning of conclave sessions was a normal oc-
currence in late-medieval and early modern times. Usually a
committee of independents-the "flying squad" (squadravolante)-
was formed to bargain with the factions (not unlike the "favorite
son" uncommitted delegates at United States party conventions).
As suggested above, the endogenous creation of candidates tended
to raise the bargaining costs.

FOCAL POINTS IN COALITION-BUILDING Given the cardinals'

penchant for factions, we would expect papal elections to have
featured numerous voting rounds, the number of candidates hold-
ing relatively stable as the rounds progressed. To test these infer-
ences, we offer interesting, albeit incomplete, data about voting
in twenty-four of the conclaves in which approval balloting was
in effect. Inevitably, our data are only as good as our sources,
which reveal that average participation was about forty cardinals
per conclave (with limit values of eighteen and sixty). The number
of candidates voted in the first round was generally large, reaching
twenty at least once.27
25 Lector, Le conclave;anon., Histoiredes Conclavesdepuis Clement Vjusqu'a present(Paris,
1689).
26 Michel Mollat du Jourdin and Vauchez, Un temps d'epreuves (1274-1449) (Paris, I990),
io8-II3.
27 Data in this section not otherwise referenced are from Histoire des Conclaves, pages too
numerous to list.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
20 | JOSEP M. COLOMER AND IAIN McLEAN

Despite the many candidates in the running, individual car-

dinals limited themselves to only a few. The mean number of
candidates nominated per voter seems to have been between I.5
and 2. Many cardinals did not have sufficient knowledge about
candidates to make more than a couple of informed choices, thus
bestowing an advantage on well-informed conspirators. Conclave
rules imposing silence and prohibiting explicit negotiations among
the cardinals, however, had their effect on such collusion. When
scrutiny showed that certain cardinals had voted for a suspiciously
large number of candidates on one ballot, angry protests from
other participants was likely to ensue, thus reducing the potential
consensual effects of approval balloting.28
The elimination of candidates often resulted from the so-
called "exclusion" vote, in which at least one-third of conclave
members publicly expressed their disapproval of a candidate. This
demonstration showed that the "excluded" candidate could not
expect a sufficient majority, even if he had won in previous
rounds. It was mainly the practice of nation-party cardinals, rep-
resenting a vestige of the emperor's former veto right in the
conclave.
Certain signals helped to channel votes into one "focal point,"
that is, a likely winning candidate. The limited information pro-
vided by successive rounds of voting could feed conjectures about
general preferences and candidates' chances, though unevenly. For
example, cardinalswho were old, infirm, and uncommitted were
particularly liable to change their minds from one round to the
next. Candidates who did not gain additional votes or dropped
votes along the way were likely to lose support from cardinals,
even if they had obtained a plurality or a majority in previous
rounds. In contrast, candidates who maintained or increased their
votes, even from a modest start, could turn out to be winners.29

28 In 1559, the secretaryof Bartolome de la Cueva, an old Spanish cardinal,privately asked

each member of a conclave to give his master one approval vote, as a gift to show the elder
that he had at least one friend. The trick was discovered when seventeen out of thirty-two
ballots included that name in their lists. Similar stratagems have been attributed to other
cardinals.See D. . C. B., Los Papas, 143-144; Lector, Le Conclave405, 622-623; Histoiredes
Conclaves,83, i6i, 240).
29 "Focal points" were first defined by Thomas Schelling, The Strategyof Conflict(Oxford,
I960), 57. For a book-length discussion of coordination in this class of electoral games, see
Gary Cox, Making Votes Count: StrategicCoordinationin the World'sElectoralSystems(Cam-
bridge, 1997).

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
ELECTING POPES | 21

After some time, voting shifts in general tended to increase.

Under the threat of privation, or illness, during conclave, cardinals
became more apt to enlarge their set of approved names. Entry
of new candidates was always possible because the rules called for
no elimination threshold. New, unexpected names might emerge
as compromises between two blocking coalitions, especially after
the exclusions of other candidates-many of them hardly men-
tioned as papal material a few hours before being crowned. Sup-
port for a candidate often snowballed. According to the available
data, a successful candidate might obtain 33, 50, or even I00
percent more support in access than in the immediately previous
voting round. At least three times, a candidate who received more
than half of the votes (on one occasion, 58 percent) lost the
election. In many cases, the final winner obtained only one-third
or one-fourth of votes in the first round. In I605, the eventual
winner, Leo XI, did not receive even a single vote.
Such results led the Church to abolish approval balloting in
162 / 622. The fact that the inefficacy of approval balloting was
due primarily to the countereffects of the rule of two-thirds and
the access round was not the reason why ecclesiastical decision
makers reestablished categorical voting. The selection of one name
per ballot simply made voting easier for the cardinals. The access
round, too, went out of style, after the I903 conclave, without,
however, altering the number of rounds per session. Yet, these
reforms served only to reduce the information made available to
the cardinals throughout the successive rounds and to reinforce
the degree of surprise in the final election.

For many centuries, the election of the pope was an occasion of

violent conflicts and schisms that weakened the Church and
subordinated it to secular powers. These conflicts later gave way
to long delays in decision making and, since the end of the
thirteenth century, to winning candidates who frequently came
as a complete surprise. The history of Church decisions about
electoral rules seems to have been driven by successive reactions
to unintended, undesirableeffects of previous decisions. The change
from unanimity rule to two-thirds qualified-majority rule was
fortunate, from the perspective of modern social choice literature,
since the two-thirds rule makes cycles and unstable decisions for

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions
22 | JOSEP M. COLOMER AND IAIN McLEAN

a limited number of candidates impossible. The trade-off, how-

ever, was the time needed to reach a final decision. Although the
delays provoked by the requirements of qualified-majority rule
were curbed somewhat by the physical and material restraints on
the cardinals while locked up in conclave, as well as by the
cardinals' prerogative to approve more than one candidate, the
rule of two-thirds was sufficiently demanding to make large agree-
ments difficult, especially in light of the vast number of factions,
the scanty information about candidates, and the restricted com-
munication between voters mandated by conclave rules. As a
consequence, the cardinals deferred the consensual decisions usu-
ally encouraged by approval balloting to later rounds, thus keeping
the voting sessions long. The intuitive signals that they developed
were an important recourse against further delay.
This article demonstrates that the successive reforms in the
rules for electing popes during the Middle Ages can be explained
as a series of rational responses to political problems faced by the
Church and by successive electors. Although the particular forms
that these developments took could not have been predicted in
advance, because they depended on certain contingencies (such
as the unusual utility function of Celestine V), the process as a
whole is illuminated from the perspective of social choice theory.

This content downloaded from 138.251.14.57 on Sun, 8 Sep 2013 14:57:47 PM

All use subject to JSTOR Terms and Conditions