University of Lund

1. Theory and Method

The purpose of this article is to examine and compare the formation of government in parliamentary multi-party systems. The investigation centres round the decision-making process which leads up to the appointment of the person who forms a government and the formation of a new government. The aim is then, primarily, to analyze what factors are relevant to the choice of the person or party whose task it is to form a government in parliamentary multi-party systems.

These factors are briefly analyzed in more detail by testing with empirical material a number of formulated principles of decision for the choice of the person or party that is to form a government. These principles of decision are derived from the basic parliamentary principle, which can be defined in the following proposition: A government must have the support of parliament or at least its tolerance. The principles of decision are closely related to the parliamentary principle.

The study of the formation of governments, with the stress on the aspect indicated above, is an important part of, and a fundamental preparation for, the analysis of a more far-reaching problem, which may be formulated in the following way. What determines, generally, the assignment of the power of government in a parliamentary multi-party system? The party which forms the government is an obvious starting-point for the assignment of the power of government irrespective of whether the future government will be a one-party government or a coalition government of the minority or majority type. The analysis is thus focussed on the fundamental part of all types of government composed of political parties in parliamentary multi-party systems.

Finally, it should be mentioned that a large part of the empirical material and
certain important methods are based on a previous study by the author.1

Principles of decision in the choice of the person who is to form a government

In order to be able to make a fairly careful analysis of what factors are relevant
to the choice of the person whose task it is to form a government in parliamentary

---

* The article is basedon a summary of my thesis, Regeringsbildningen i flerpartisystem (Government Formation in Multi-Party Systems), Lund, 1975.

* The article is basedon a summary of my thesis, Regeringsbildningen i flerpartisystem (Government Formation in Multi-Party Systems), Lund, 1975.

* The article is basedon a summary of my thesis, Regeringsbildningen i flerpartisystem (Government Formation in Multi-Party Systems), Lund, 1975.

multi-party systems, it is necessary to start by describing more precisely the
theories and methods on which this article is based.

To begin with, two general prerequisites for the following discussion will be mentioned. In the first place, there is general identity between the person whose task it is to form a government and the party whose task it is to do so. Secondly, there is complete party cohesion. The second prerequisite means that every party is regarded as an actor carrying the weight indicated by its current number of seats.

It is also clear from what has been said above that the investigation is restricted to a discussion of parliamentary multi-party systems and that the principles for the choice of the person or party whose task it is to form a government are related to the basic parliamentary principle that a government must have the support of parliament or at least its tolerance. This principle raises three important questions which will now be commented on in more detail: What does this principle mean more exactly? (The meaning of the principle.) Can the principle be applied not only to governments but also to parties forming a government? (The range of the principle.) Finally, in what way can the meaning of the parliamentary principle be operationalized in the analysis of the choice of government-forming party? (The operationalization of the principle.)

By way of introduction it may be said about the first question that the meaning is comprehensive in the sense that several observable states (support or tolerance) can occur when a government is being formed. In this connection there is reason to try to determine the two states that indicate the upper and lower limit to (or the maximal or minimal meaning of) the parliamentary principle.

The upper limit may be defined as the situation where a new government has the support of all the parties in the sense that all the parties are represented in the government. In accordance with the assumption that party cohesion is complete, it then follows that the government is supported by 100 % of the seats. The lower limit may be defined as the situation where the new government does not have a majority against it. In this case the support may be confined to embracing the party or parties that make up the new government.

In the mid-positions between the upper and lower limit there are thus a number of variations in the degree of support and/or tolerance, e.g. the two main types of parliamentary representation for governments in parliamentary multi-party systems, namely majority and minority governments.

As for the second question, the one concerning the range of the parliamentary principle, it may first be established that the parliamentary principle refers to a government's support in parliament. The relation between the parliamentary support of a government and that of a party whose task is to form a government can then be described in the following way. The government-forming party is one of the primary components of all governments consisting of political parties, irrespective of whether the future government will be a one-party government or a coalition government of the minority or majority type. From this it follows that the choice of the government-forming party takes place by the application of the parliamentary principle.

This principle, however, can be applied in several ways. In order to make it possible for us to analyze its application more closely, the parliamentary principle needs to be specified and operationalized. These operationalizations must then have the common characteristic that, in various ways, they become signs of the existence of such a parliamentary support that a party (in the situation in question) can be entrusted with the task of forming a government.

The operationalizations are formulated as hypothetical principles. They are not hypothetical in the sense that they are formulated like conditional clauses. But, first of all, they are hypothetical in the sense that their applicability to the empirical material is tested from one case to another. They are also hypothetical in the sense that they are looked upon as principles with a potential explanatory power - a rational reconstruction - and not as statements as to what psychological factors have influenced the decision of the various actors in one case or another. The principles thus form starting-points for an explanatory theory of why and how a government is formed. For such explanations to be tested, one must of course require that it should be possible to establish whether the operationalizations are applicable or not in the respective cases.

In this study six operationalizations of this kind are formulated. Their mutual relations and their role in the decision-making process will be treated in more detail in the next section. The operationalizations may be expressed in the following propositions:

(1) Existence of two-way possibilities of co-operation2

(2) The party is the largest party

(3) The party has obtained the greatest increase in mandates

(4) Existence of a majority recommendation

(5) Existence of a bloc majority

(6) The party plays the main role in bringing about the fall of the resigning

government.

What has been said may be summed up and explicated in the following manner. Thus, in the first instance, one ought to make a distinction between the parliamentary principle, various operationalizations of this principle, and the circumstances which are set forth in these propositions. Secondly, the decision to entrust someone with the task of trying to form a government is taken by the head of state. We can then distinguish between

(1) The head of state's decision to entrust X with the task of forming a govern

ment, and

(2) The reasons that have been invoked or could be invoked for this decision.

For these reasons to be relevant there ought to be some general pattern of the
following type:

(a) If conditions A, B, and C are present, the head of state will entrust (alternatively:
the head of state ought to entrust) to X the task of forming a
government.

One of the main aims of this investigation is to find out if and to what extent it
is possible to discern any general patterns of this kind in historical material and,
if so, what characterizes these patterns.

Thus here I intend to use the parliamentary principle as a point of departure, propose a few operationalizations of it, and test the hypothesis that these operationalizations, individually or in certain combinations, are used as bases for the decision to entrust the task of forming a government to a particular actor. If and to what extent this is the case can be determined only by scrutinizing empirical material, which will be commented on more fully below. Since these operationalizations (and the circumstances set forth in them) are closely connected with the head of state's decision, and thus play an important part in the decision process, they will from now on be called decision principles. However, the terminology is not very important in this connection; what is important is to see how these operationalizations relate to the parliamentary principle and to propositon (a) above. The operationalizations must not be confused with this proposition. The circumstances set forth in the operationalizations are circumstances of the kind described in the //-clause in (a).

The decision principles and the decision process

In this section the mutual relations of the decision principles will be commented
on in more detail as well as their function in the decision process.3 The actors in
the decision process and their respective roles will also be described more precisely.

By way of introduction, it was stated that the decision process up to the choice of the party whose task it will be to form a government is in the forefront of the investigation. The natural conclusion of the decision process consists in the government-forming party carrying out its task, i.e. forms a government. The startingpoint of the process can then be considered analogous with the situation in which the government in office decides to resign or is forced to do so and hands in its resignation. At this stage the process that leads up to the formation of a new government may be said to begin.

Of course, the exact time for the starting-point of the process may be discussed and even called in question considering that various types of preparations may take place, for instance negotiations between parties before the government hands in its resignation. Nevertheless, it may be said that such preparations somehow become evident or emerge in the ensuing process of decision. Thus no important aspect of the choice of government-forming party is likely to be neglected by connecting the starting-point of the decision process with the situation where the government in office hands in its resignation.

After having established the starting-point and the conclusion of the decision
process we should look a little more closely at the actors that take part in the

decision process and their respective role in that process. In agreement with the above-mentioned purpose in this article, the parliamentary parties are the chief actors in the decision process. It is taken for granted that the leaders of the parliamentary parties represent all parliamentary members of their respective party. The role of the parliamentary parties in the decision process may be described briefly as the 'subject' and 'object' of the process.

The 'subject' part of the process comprises two important factors, i.e. a calculation of a party's own interests⁴ involving in the first place the desirable solution of the question to whom the assignment of the task of forming a government should go, and secondly, the question of the political composition of the government. The second factor may be expressed as a combination by the party of its own interests with an assessment of a desirable or 'reasonable' solution of the question of what party should form a government with regard to the meaning of the parliamentary principle. Every parliamentary party functions as the 'object' of the process in the sense that other parties assess the position and suitability (in a broad sense) of the party when it comes to the assignment of the task of forming a government.

Another actor in the decision process is the head of state,⁵ whose duty it is, in the political systems dealt with in this paper, to assess the import of the recommended choice of the parliamentary parties as to what party should be given the task of forming government. Thus, within the scope of the parliamentary principle it is up to the head of state to make decisions as to the assignment of the task of forming a government.

Apart from the parliamentary parties and the head of state there may be
advisers to the head of state who take part in the assessment of the situation which
precedes the decision to select the party that is to form a government.

After the above discussion of the framework of the decision process and the
actors taking part in that process, the mutual relations and function of the decision
principles in the decision process will be detailed below.

The parliamentary principle has been operationalized in six different manners. These six different types of parliamentary support may be divided into two main categories. First, we have the number of seats and the increase in the number of seats which are represented by the decision principles 2 and 3 and partly by 5. By 5 is here meant that a party together with other parties has a majority at its disposal.

Secondly, there is a specific way of determining a party's position in relation to other parties in the system. This is represented by the decision principles 1 and 5. Principle 5, in this case, refers to the determination of the relative positions for a group of parties.

Concerning the decision principles 1, 2, 3, and 5 mentioned above, we may observe that they are referable in time to the beginning of the decision process. Thus, the potential basis for the applicability of the above-mentioned decision principles is at hand at the start of the decision process.

However, it may be noted that the operationalization of decision principle 6 is
such that it precedes or triggers the subsequent decision process. Finally, decision

principle 4 is entirely related in time to the material contents of the decision process in the sense that its potential applicability to a party becomes evident from the attitudes and standpoints of the actors (in this case the parties) during the decision process.

2. Selection Principles for the Investigation

In the selection of the area of investigation various viewpoints have been considered in order to carry through a five-step selection. The considerations that have guided the selection of government formations are described in more detail under the following headings:

(1) Countries

(2) Number of government formations

(3) Period of investigation

(4) Spread in tune within the period of investigation

(5) Variation in the causes that bring about the resignation of a preceding government.

Countries

It was earlier stated that the investigation deals with parliamentary multi-party systems. With this limitation the following six minor European multi-party systems have been selected: Belgium, Denmark, Finland, the Netherlands, Norway and Sweden.

Number of government formations

The investigation centres round the decision process that leads up to the appointment of the person who is to form a government and to the formation of a new government. It is evident that a condition for such a decision process is that the government in office hands in its resignation. For this reason six government formations have been chosen in each country except Sweden (see below). The investigation comprises a total of 41 government formations.⁶

Period of investigation

The government formations examined must have occurred during the last 10 to
15 years. The period at the end of 1973. However, Denmark and Sweden are in
certain respects exceptions to this rule (see below).

Spread in time within the period of investigation

The government formations examined must be as far apart in time from each other as possible. This stipulation was made in an attempt to prevent particular circumstances of one kind or another from influencing the decision process and the choice of the party whose task it was to form a government.

Variation in the causes that bring about the resignation of a preceding government This criterion is similar to criterion 4 in that it stipulates that the selection of causes which bring about the resignation of a preceding government should be as varied as possible.

The idea is thus that the ensuing decision process and the choice of governmentforming party must not be entirely dependent on the cause of the resignation of the government in office. With respect to this criterion, Denmark has been made an exception as far as the period of investigation is concerned, which in this case has been extended to cover 20 years. If this had not been done, nothing but defeat in general elections in all six cases would have been the cause of the resignation of the government in office.

In order to supply various causes for the resignation of a government the following
factors have been given due consideration:

(a) Defeat in a general election

(b) Defeat in parliament on an issue vital to the existence of the government
(c) Rift in a coalition government

When it comes to the respects in which Sweden is an exception the following remarks may be made. To be able to cover the six latest government formations it is necessary to go back 43 years in time. The definition of government formation mentioned above applies also to this point in time. Since the time period is markedly different from that used for the other countries there is reason to treat Sweden differently. That is why all the eleven government formations in Sweden after 1921 have been dealt with. That year has been chosen as the starting point because universal and equal suffrage was introduced in that year.

As far as Norway is concerned, the government formations accounted for are the six latest within the time period 1963-1973. From 1945 to 1963 one party (Arbeiderpartiet) was in office. Besides, that party had a majority of its own in parliament (stortinget) from 1945 to the election of 1961.

3. Comparison between the Decision-Making Processes during the Formation of Governments

A comparison based on the selected government-formations in the countries chosen might reveal if and in what respects the decision-making processes differ. This comparison includes the behaviour and role of the head of state in the abovementioned countries.

In most cases the process starts by the existing government handing in its resignation. But the case of Finland is, in some respects, rather different. Under certain circumstances the Finnish president has taken the initiative in starting negotiations with representatives of the different political parties before the government handed in its resignation.

The order in which the representatives of the political parties have opportunities
to influence the head of state is more or less dependent on a formal decision

taken by the head of state. Well-established custom restricts the scope of decision for the head of state. This applies to Finland, Belgium, and the Netherlands; even in Denmark we have the same situation, which will be shown towards the end of our period of investigation. The well-established customs mentioned above are in all cases the same: the representatives of the political parties, beginning with the members of the largest party, then those of the second largest party, etc., are called to the head of state for negotiations. The exception to this rule is Sweden, where this principle has not been applied except during the government-formation of 1957.

What political parties are called to the head of state is also regulated by custom. In all countries except Belgium and Sweden all parties represented in parliament are invited to take part in the first round of negotiations presided over by the head of state. In the latter countries the small parties are excluded, at this stage, from the opportunity of giving their recommendations to the head of state. But in negotiations that follow beween the parties and the head of state no general pattern can be observed. The current political situation and the cleavages between the parties decide which parties are to take part in the negotiations and also the duration of the period of negotiation.

In the contacts between the parties and the head of state the parliamentary leaders represent their parties. This goes for all countries except Belgium, where all the political parties are represented by their chairmen. The important thing is, however, that the parties decide which person(s) are to represent them. It has thus been shown that the head of state has not tried to interfere in internal party rifts by choosing a particular party representative.

The representatives of the parties inform the head of state about the opinion of their respective party concerning the desirable composition of the government to be. At the beginning of the decision-making process the recommendations given by the party-representatives are often vague and lack explicit information concerning who is to be assigned the task of forming the new government and which parties should be included in the new government.

The nature of the recommendations at this stage of the process could be explained by the desire of the parties to maintain freedom of action in politically unclear situations where different solutions are possible. The coming negotiations between the parties should not be hampered by declarations which are so specific as to restrict the choice to only one of many possible solutions. As the decisionmaking process develops, the recommendations become more and more distinct as to what person is to form the government and it also becomes clearer what kind of parliamentary basis there will be in terms of political parties for the new government.

In addition to the advisers mentioned earlier the head of state always consults the speakers in parliament, except in Norway and Denmark. Furthermore, there is a certain freedom for the head of state in the Netherlands and Belgium to choose advisers outside parliament and the political parties.

For the ongoing assessment of the recommendations during the above-mentioned
process, a predominant role is played by the person of the head of state in all

countries except Norway. In this country the negotiations between the parties have in fact replaced the assessment made by the head of state. However, in order to carry out the important function during this process the head of state is entitled to seek special assistance from outsiders.

In Denmark a civil servant at the court (kabinetssekretaer) acts as a permanent adviser to the head of state, with the task to assist in the evaluation of the recommendations. In Belgium and the Netherlands a special representative mediator (informateur) usually both presides over the negotiations between the parties and acts as a kind of assistant head of state when it comes to assessing the situation. In the choice of the person who is to be given the task of forming a government the head of state can personally influence the situation. If, during the whole decision-process, the recommendations have been vague and unspecified with respect to a particular candidate, the result is that the head of state will be forced to put his own interpretation on the contents of the recommendations and on the attitudes shown by the parties before a new government is formed. In these situations the heads of state have a possibility to influence the final choice of a person.

The decision to entrust a person with the task of forming a government can be formulated in different ways. Most tasks are simply stated without any further qualifications, namely just 'form a government'. But in all countries, except the Netherlands, there are examples of conditions applied to the tasks of forming a government. In the Netherlands a certain standard formula is used in all cases. But for the other countries four main types of conditions have been used on different occasions:

(a) All-party government
(b) Majority-government

(c) Bloc-government (e.g. the non-socialist parties)
(d) Explicit naming of government-party or parties

Obviously there are possibilities for the head of state to influence the task of forming a government, e.g. either to promote and/or avoid certain types of government. However, it should be observed that every condition is not intended to influence the composition of the future government. In many cases they are based on agreements reached by the political parties during the preceding negotiations.

In order to sum up the differences in the role of the head of state in the investigated countries, we may conclude that the main differences are between Finland and the other countries. The Finnish president has on several occasions taken personal initiatives at an early stage of the process. He has also actively taken part in the negotiations between the parties in order to promote certain agreements between the parties in order to arrive at a possible basis for a future government. He has also applied different conditions in the tasks of forming a government to promote a desirable composition of the future government. What has been said applies to political parties and/or to the parliamentary basis for the government.

Finally the President of Finland has supported the person who has been chosen to
form a government in different ways in order to make it possible for him to carry
out the task of forming a new government.

4. Summary of Conclusions

The operationalization of the parliamentary principle in six decision principles has proved to be useful in an analysis of what factors determine the assignment of the task of forming a government to political parties in parliamentary multi-party systems. The decision principles have been applicable to 90 per cent of the tasks of forming a government that were examined. Although, of the six decision principles, decision principle 6 proved to be of limited value, in an investigation of all factors it did not seem wise to exclude this decision principle. As for the details of the assignment of the decision principles applicable to various government formations the reader is referred to the appendix showing government formations, the tasks of forming a government, the combination of decision principles, the government-forming party, majority governments, and coalition governments.

The decision principles were supposed, by definition, to function as indicators of various forms of parliamentary support for a party. An analysis of the degree of the mutual relationship of the decision principles showed that they did not overlap or cover each other to such an extent that any decision principle could be excluded. This should be stated with reservation for decision principle 3, which to a large extent is covered by decision principle 2. An additional check was undertaken to determine if in their application the decision principles indicated various forms of parliamentary support rather than exclusively referring to the size of the party. An analysis showed that the number of applicable decision principles could not be said to have any clear cumulative effect on party size.

After a more detailed description of the functions of the decision principles in the decision process, such as functions of size and co-operation, these functions were related to party size. The analysis then showed that the shortcomings in size properties of a party could be compensated for by more prominent properties of co-operation in that party. Thus, the sufficient explanation that every combination of decision principles may be supposed to contribute to the assignment of government-forming tasks can be composed in various ways.

When all the factors influencing the assignment of tasks of forming a government have been considered, a conclusion can be drawn as to the possibility of compensation for deficient size properties by virtue of prominent properties of cooperation. Such a conclusion should be made basic to a hypothesis which aims to specify the limits to this possibility of compensation.

Inherent in the argument above is the conclusion that a party, in order to be
assigned the task of forming a government, must have both properties of size and
properties of co-operation.

In connection with a comparison between potential and factual tasks of forming
a government, tasks that did not get any marking or only one marking were also
examined. The analysis in that section led to the conclusion that the difference, by

and large, between potential and factual tasks of forming a government was that the parties that were assigned potential tasks were deficient in both properties of size and properties of co-operation. The possibility for a party to realize tasks - factual tasks that is - of forming a government increases with the number of applicable decision principles. Furthermore, the parties that have been assigned factual tasks are more prominent in size properties as well as in properties of co-operation according to the decision principle called majority recommendation. From tfiese facts the conclusion could also be drawn that two-way possibilities of co-operation indicate weaker properties of co-operation than majority recommendations with regard to the possibility for a party to realize any task of forming a government.

To sum up, the conclusions which the investigation draws have provided the
basis for the following preliminary explanatory model for the assignment of tasks
of forming a government in parliamentary multiparty systems.

(Appendix) Government Formations, Tasks of Forming a Government, Combinations of Decision Principles, Government-Forming Party, Majority Governments and Coalition Governments

(Appendix) Government Formations, Tasks of Forming a Government, Combinations of Decision Principles, Government-Forming Party, Majority Governments and Coalition Governments

(Appendix) Government Formations, Tasks of Forming a Government, Combinations of Decision Principles, Government-Forming Party, Majority Governments and Coalition Governments

1 Reference is here made to a task that was assigned to the two parties, both of which had
two-way possibilities of co-operation.
2 Non-Socialist majority.
3 Socialist majority.
4 'White collar' government.
5 First letter(s) in a series, e.g. sa, 7a-7c, denote(s) a task that was not carried out (potential),
whereas the last letter in the same series, e.g. sb, 7d, denotes a task that was carried out
(factual).
6 T = Sufficient cause for being assigned the task of forming a government.
7 Largest opposition party.
8 Non-Socialist majority.
9 Socialist majority.
10 Largest bloc party.
(P) = Axel Persson-Bramstorp.
(H) = Per Albin Hansson.
11 T = Sufficient cause for being assigned the task of forming a government.
12 The combination is basedon PvdA, which was assigned a task in collaboration with ARP.

Party labels
Denmark Senterp. = Center Party (agrarians)
Soc.d. = Social Democrats Krist.fp. = Christian Peoples Party
RV = Radical Liberals
V = Agrarian Liberals Sweden
Soc.d. = Social Democrats
Finland Nat.p. and Lantm.p. = Conservatives
Folkd.d. = Communists Fris.fp. = Christian Peoples Party
Nat.s. = Conservatives Bondef. := Farmers Party
Agrarp. = Agrarian Party
Fi.fp. = Finnish Peoples Party Belgium
Sv.fp. = Swedish Peoples Party PSC = Christian Socialist Party
Soc.d. = Social Democrats
The Netherlands
Norway KVP = Catholic Peoples Party
Høyre = Conservatives ARP = Anti-Revolutionary Party
Arb.p. = Social Democrats PvdA = Social Democrats

---

NOTES

1. Henrik Hermerén, Regeneringsbildningen i flerpartisystem (Government Formation in Multi-Party Systems) (Lund: 1970). mimeographed.

2. The operationalization 'existence of two-way possibilities of co-operation' is in short, whenever applicable, primarily basedon the position of the party on a co-operation scale. The principle behind the construction of co-operation scales for the chosen countries can be illustrated by the following diagram: Suppose that a coalition has existed betwen A and B, between B and C and between B and D (1) A-B (coalition 1) t* (2) B-C (coalition 2) t2 (3) B-D (coalition 3) V Then co-operation possibilities could then be illustrated by connecting the letters (parties) in the following way: A B<^^ B has obviously a possibility, according to this two-dimensional co-operation scale, to cooperate with, for instance, C or D (given that certain assumptions are fulfilled, e.g., that the order of preferences t¹ and t3t³ of the parties are not significantly changed). Suppose further that the coalition BCE has existed. To the given information (1) - (3) we could then add: (4) B - C - E The co-operation possibilities could then be represented in the following diagram: A *X<^^ Given the information (1) - (4) there are co-operation possibilities between, for instance, A-B-C and B-C-E but not for A-B-E. For this reason E must be directly connected with C and not with B. From the diagram we can further see that if A is to co-operate with three parties this must first include B and then C or D; and if A is to co-operate with C or D, B must also be included. Such a prediction is basedon the assumption that the relative ranking order between th parties as it has been manifested by the coalitions (1) - (4) is relevant at the time of the prediction. In the diagram we can also see that the parties B and C have two-way possibilities of co-operation along a horizontal dimension. Finally, it is of course the earlier co-operation structure between the parties that is taken into account in the selected government formation.

---

2. The operationalization 'existence of two-way possibilities of co-operation' is in short, whenever applicable, primarily basedon the position of the party on a co-operation scale. The principle behind the construction of co-operation scales for the chosen countries can be illustrated by the following diagram: Suppose that a coalition has existed betwen A and B, between B and C and between B and D (1) A-B (coalition 1) t* (2) B-C (coalition 2) t2 (3) B-D (coalition 3) V Then co-operation possibilities could then be illustrated by connecting the letters (parties) in the following way: A B<^^ B has obviously a possibility, according to this two-dimensional co-operation scale, to cooperate with, for instance, C or D (given that certain assumptions are fulfilled, e.g., that the order of preferences t¹ and t3t³ of the parties are not significantly changed). Suppose further that the coalition BCE has existed. To the given information (1) - (3) we could then add: (4) B - C - E The co-operation possibilities could then be represented in the following diagram: A *X<^^ Given the information (1) - (4) there are co-operation possibilities between, for instance, A-B-C and B-C-E but not for A-B-E. For this reason E must be directly connected with C and not with B. From the diagram we can further see that if A is to co-operate with three parties this must first include B and then C or D; and if A is to co-operate with C or D, B must also be included. Such a prediction is basedon the assumption that the relative ranking order between th parties as it has been manifested by the coalitions (1) - (4) is relevant at the time of the prediction. In the diagram we can also see that the parties B and C have two-way possibilities of co-operation along a horizontal dimension. Finally, it is of course the earlier co-operation structure between the parties that is taken into account in the selected government formation.

3. For a survey of various types of decision theories the reader is referred to e.g. James A. Robinson & R. Robert Majak, 'The Theory of Decision-making', in James C. Charlesworth (ed.), Contemporary Political Science (New York: 1967), chp. 10.

4. For a more detailed discussion of the calculations of the parties the reader is referred to, e.g. G. Sjoblom, Party Strategies in a Multiparty System (Lund: 1968).

5. See, e.g. J.-F. Lachaume, 'Le role du chef d'État dans les monarchies parlementaires d'aujourd'hui', Politique 41 (4), 1968, pp. 87 ff.

6. We get an idea of the size of the selection by comparing it with the total number of governments between 1945 and 1970 in the countries chosen. The number of governments during this period amounts to 76. This figure is based on a compilation in K. von Beyme, Die parlamentarischen Regierungssysteme in Europe (Munich: 1970). It should be noted that von Beyme also considers minor changes in a government formation to constitute a new government. Thus, according to von Beyme, the government in office is not required to hand in its resignation for it to be regarded as having constituted a new government.