CHAPTER 2 - GENERAL SYSTEMS THEORY
Copyright [c] 1986 by Allan G. Feldt
Beginning during the 1930's and accelerating after World War II, scientists and philosophers from a number of different disciplines began to publish and discuss a series of papers on the common properties found in all systems. This search for universal laws pertaining to all systems came to be called General Systems Theory. It has borrowed heavily from a number of disciplines but has also made important contributions to them. Among those most effected are: Biology, Chemistry, Computer Science, Economics, Information Theory, Operations Research, Philosophy, Physics, Psychology, and Sociology.
Full coverage of the substantial literature on General Systems Theory is well beyond the scope or purpose of this book. However, a few basic concepts found in General Systems Theory have proven to be especially useful in increasing our understanding of urban and regional phenomena. Nine of these basic concepts will be presented here in order to lay a coherent groundwork for the diverse theories and ideas borrowed from a wide variety of social science disciplines which make up the remainder of this text.
1. Definition of a System
A system may be defined as any set of two or more interdependent parts which has a relatively high degree of closure, connectivity, and stability
Under such a simple definition, virtually any pair of persons or objects can be considered as a potential system since at some level of specificity it can usually be demonstrated that "everything is related to everything else." Clearly such a minimum definition is almost useless without the three important qualifiers provided at the end of the definition, i.e., "a relatively high degree of closure, connectivity and stability." Indeed, it is these qualifiers which make this particular definition of a system especially useful by providing the criteria needed to evaluate the quality of the system being considered.
Once understood, these three criteria may be applied to any system at any level to determine whether it is worth further consideration as a system. In each case the observer must examine a potential system and judge its performance on these three criteria relative to other systems of similar type or other possible combinations of elements on which attention might alternatively be focused. Based on personal judgement and such measures as are available, the observer must then decide whether or not to continue study and analysis at this level. That choice is made personally and professionally and is not inherent in the system being considered. The meaning of these three qualifiers is discussed further below.
A) Connectivity - the degree of internal interdependence of a system - is reflected by exchanges occurring between the parts making up a system. If two or more parts of a system are highly interdependent, they must engage in a large number of interactions. If no interactions occur between the parts of a system, they are not interdependent and therefore they do not make up a system. Interactions may be made up of material goods, information, energy, services or even intangibles such as love and affection. Clearly some types of interchanges are more important than others and their volume and frequency of occurrence must also be considered. Direct measurement and comparison of interactions is extremely difficult in most situations, however.
If we assume for the moment that all types of interactions are of equal significance, we may define the degree of connectivity of a system as the proportion of all those interactions which occur within the system compared to the maximum number of interactions which could theoretically occur. Put another way, connectivity is equal to the number of interactions which both begin and end within the system divided by the theoretical maximum number of interactions which could begin and end within the system.
The denominator in such a proportion is determined theoretically in terms of the number of separate parts the system contains and the number of kinds of interchanges each can engage in. The numerator is simply some measurement or observation of the number of interactions which occur over a reasonable period of the system's operation. While sometimes difficult to measure, the concept is clear and a variety of indirect measurements of such interactions can be found. Clearly such a proportion varies from a minimum value of zero for non-systems to a maximum value of 1.0 or 100% for systems where all possible interactions occur, i.e. in a "totally connected system."
Consider the following examples illustrated in Figure 2.1 of possible systems each made up of four components where a one-way interaction is indicated by a single-headed arrow and a two-way interaction is indicated by a doubleheaded arrow. Counting two-way interactions as
two interactions, the maximum number of interactions possible among four components is twelve. Assuming all interactions to be equal in value, the degree of connectivity of each system is as indicated.
System I is a simple one way causal chain which is relatively rare except under artificial conditions. Such a linkage might be characteristic of some military or work groups where strict obedience to commands is necessary and the opinions of "lower" members of the system are irrelevant. It might also characterize certain food chain relationships at least during periods of abundance. In a simple predator/prey chain humans hunt foxes, foxes eat rabbits, and rabbits eat grass. If such relationships described a family, however, it would probably be a weak, unhappy, and potentially unstable system. Similarly, the amount of grass available probably has direct effects on human beings as well as on rabbits and foxes.
System II illustrates a complete loop plus two additional one-way connections between system members. Such a system might again characterize certain military or work-group systems where A and B each give orders to two other members but only receive information or material from one member. The leader of the group is probably A, since it initates interaction with B and C while B only initiates interaction only with C and D. In a food chain such a system would describe one in which humans hunt both foxes and rabbits, foxes eat both rabbits and grass, and the availability of grass affects the well being of humans, possibly providing forage for their horses. If describing a family, this system appears a little better but still quite "weak" with no members responding directly to the initiatives of others.
Systems III and IV are much more highly connected, illustrating rather complex chains of military or work group organization or a highly intricate food chain with multiple and two way relationships between most members. Note, however, that even the most highly connected system illustrated, IV, is still a little weak when used to illustrate a family, one member still not responding to one other family member. Highly connected systems involve very complex patterns of interactions among system members. Despite their complexity, hwoever, all of us have had substantial experience with many types of highly connected and complex social systems such as families, work groups, clubs, universities, etc.
B) Closure is a similar concept with the same measure of interactions in the numerator, i.e. the number of interactions which occur among the parts of the system. In this case, however, the denominator is a measure of all interactions in which any component of the system is involved, including interactions with components outside the system. Put another way, system closure is the number of interactions which both begin and end within the system divided by the number of interactions which either begin or end within the system. As with connectivity, the value of closure may vary from 0.0 to 1.0, the former for a totally open system in which all interactions occur with components outside the system and the latter for a totally closed system engaging in no interactions with outside components.
Once again, the actual measurement of interactions is difficult and the simplifying assumption that all interactions are of equal value is required for illustrative purposes. Using the same four systems illustrated above but now allowing a specific number of "outside" interactions to also be counted, the following levels of closure may be computed.
Systems I through IV in Figure 2.2 illustrate the same food chain, military, or work group examples discussed in relation to Figure 2.1 except that the interactions of individual components with other components outside the system are now also considered. In these systems, A is involved in four external interactions, three of which are incoming, while C receives one external interaction and D initiates two. In a military system D might represent persons actually firing at the enemy while A would be the officer in charge and B and C would be support personnel. In a work group D might represent a shipping clerk while in a food chain D might represent a commodity such as water or shade whose supply or condition is important to the well being of several components outside the chain being considered.
All four of these systems have seven outside contacts in addition to those occurring internally. The closure of each successive system increases since the only new interactions being added are those occurring totally within the system. System I is quite open with system components involved in more external interactions than internal. Such a system might be expected to be fairly weak and potentially unstable, although this conclusion is not necessarily correct in all circumstances.
In contrast, System IV has more internal than external linkages and might be considered a rather strong system. Such terms can be misleading, however, since in a chain of interdependent parts all components are equally important and the total configuration is as weak as its weakest component.
It should also be noted that external interactions can vary independently of the number of internal interactions. Thus a system with high connectivity may also have low closure. Indeed, this is true of most current urban societies which have many internal interactions but which are also highly dependent upon the rest of the world. Figure 2.3 illustrates this condition for a System V which has the same number of internal interactions as System IV but many more external interactions.
Before completing our discussion of closure and connectivity, it is worth mentioning two subsidiary concepts from general systems theory which are illustrated in the examples just reviewed. The first is that when dealing with interactions external to the system under consideration, it is not necessary to specify the external component with which the interaction is occurring. It is sufficient to know that the interaction begins or ends outside the system. If it becomes important to identify the external components with which the interaction is occurring, the nature of the study itself has been shifted so that a new, larger and more inclusive system must be defined and evaluated for connectivity, closure, and stability.
The second concept illustrated here is that of dominance or power. In general, initiating interactions is an indication of power or influence over other components while receiving interactions indicates a position of lower influence. Thus when considering only interactions internal to the system as in Figure 2.1, component A appears to be generally dominant since it initiates interactions for other members of the group including B. In contrast, D appears to be the most subservient component in Systems I, II, and III since it is the recipient of many interactions, initiates few, and lies at the end of the chain of interaction flowing from A to B and C. In System IV, C and D appear to be equal in that they receive and initiate equal numbers of interactions while A still dominates by initiating one more interaction than it receives.
This generalization must be treated cautiously, however, particularly in the case of interactions with components external to the system. Components with many interactions outside the system are in a "gatekeeping" role for the rest of the system and often become the most influential components within the system. Whether such external interactions are receiving or sending exchanges must also be considered. System components which receive more interactions from external sources are usually more influential within the system while system components which initiate more interactions to external components are usually more influential outside the system.
C) Stability, the third definitional property of a system, refers to the relative length of time the system exists or recurs in substantially the same form. Since most systems are in continual change and evolution the interpretation of stability is subject to considerable variation depending upon the needs of the particular observer.
Furthermore, what constitutes a significant length of time varies greatly both among observers and among various types of systems. When dealing with subatomic particles, milliseconds are appreciable periods of duration for some systems. Traffic jams and theater crowds have durations of a few hours. Cities and counties exist over decades and centuries while civilizations and geological changes occur over millenia. The length of time during which the system exists must be sufficient to allow adequate observation and to allow internal self-regulating mechanisms to function. The length of time required, however, must be evaluated in relation to other systems of the same type - not against any absolute standard.
Before concluding our discussion of the definitional properties of systems, a few unproven generalizations on the nature of closure and connectivity may be of interest:
a). Closure tends to be greater for large systems than for small systems. This is a natural occurrence since expanding the size of a system to include additional components usually, but not always, transforms a number of external interactions into internal interactions.
b). Connectivity tends to be less for large systems than for small systems. This is also a natural occurrence since increasing the number of components in a system increases the number of possible interactions which could occur within the system exponentially.
c). Each type of system has a characteristic degree of closure, connectivity, and stability and other systems of the same type should be compared to that characteristic level. This consideration is behind the use of the phrase relatively high degrees of closure, connectivity, and stability.
d). The longer a system can survive when totally cut off from all external relations, the greater its degree of closure.
e). The faster a system is effected by the loss of one of its components, the greater its degree of connectivity.
2. System Development
The growth and development of a system is usually characterized by increasing specialization of system components and increasing exchanges between the system and other components outside the system. In the terms used earlier, developing systems tend toward higher levels of connectivity and lower levels of closure. Systems with more highly specialized components and higher connectivity are often referred to as being more "complex." Systems with more external relationships are referred to as being more "dependent" or less "isolated." The evolution of human civilization is largely a story of such development, i.e., increasing specialization and trade within social systems and increasing levels of external contact through trade, exploration, and military conquest.
Consider an island made up of four small villages: A, B, C, and D. For the sake of simplicity let us assume that the basic needs of the islanders are met by four commodities: rice, fish, bananas, and building materials such as palm leaves and timber. At its least developed state, each of the four villages may satisfy all the needs of its own residents and no economic exchange occurs among the villages or between the island and the outside world. Considering the island as a system and the four villages as its component parts, at this stage of development connectivity is zero and closure is 1.0. In fact, the island is not a system since there is no interdependence among its parts.
It should be noted in passing that even when there is no economic advantage to exchanges between villages, social and political interactions are still likely to occur, sometimes accompanied by ritualistic exchanges of commodities in what appears to be an economic exchange. Such activities help to maintain alliances and friendships thereby providing a basis for resolution of disputes, defense against outsiders, and possibly intermarriage between villages. Malinowski (1961) and several other anthropologists have described such "economically irrational" trading systems. Similar economically irrational patterns are found in the exchange of Christmas gifts and dinner invitations in western societies.
When variations in rainfall, soil, minerals, altitude, or local skills produce differences in the quality and quantity of the commodities produced in different villages, however, it is to everyone's advantage to trade commodities between villages in order to take advantage of the higher productivity or better quality of goods produced by the favored village. Thus one village may have better soils for rice cultivation, another may be nearer prime fishing areas, and so on. Given even a slight advantage in producing certain commodities at one location or another, trade among villages is likely to emerge and the connectivity of the system increases. In a minimally connected system like System I in Figure 2.4, perhaps only one village would specialize in one commodity, exchanging it with the other villages for whatever items they might be able to offer.
At a somewhat higher level of development, each of the four villages might specialize in one of the four commodities, producing a fairly highly connected system within the island but no external trade. At a still higher level of development one or more of the villages may be engaged in trade with outside groups for commodities not readily available on the island, such as pigs or milk. In this case, the specialization of villages is likely to increase further and all of them would produce some surplus commodities to be traded for goods received from outside the island, possibly through a single village acting as intermediary. Such a system is much more fully developed and usually has a higher standard of living than the earlier examples, at least in terms of material well-being.
More developed systems with higher degrees of specialization and trade, both internally and externally, are generally more productive and provide stronger, more secure livelihoods for their members. The transition of any society from a less developed condition of low connectivity and high closure to a more developed condition of high connectivity and low closure is commonly viewed as "progress." Some social theorists question that assertion, however.
3. System Control
As the connectivity of a system increases the system becomes increasingly vulnerable to disruption due to the failure of one of its specialized components or breakdown of one of the interactions. This tendency is well recognized and most complex systems devote part of their resources to protecting against such disruptions. There are five methods of preventing or minimizing such disruptions.
a). Administration: The most common defense against system breakdown is through devoting part of the resources of the system to administering and controlling the activities of other system components. By providing more supervision, more regulations, more planning, and increased numbers of personnel in administrative positions, systems attain better control over their own activities and are able to anticipate and prevent interaction and production problems among their components.
Such administrative proliferation is well known and is reflected in the popular literature by Parkinson's Laws on bureaucratic growth. Studies of large organizations reveal that as the organization grows in complexity, an ever larger proportion of its employees become involved in administrative activities rather than in actual production. It is not increases in the size of an organization which results in increased administrative activities, however. Rather it is increases in its complexity, i.e. increased connectivity. Since greater size and increased complexity often occur together, it is common to attribute increases in administration to increased size rather than to increased connectivity.
While larger and more centralized administration may help a social system to operate more smoothly with fewer disruptions and greater equity for all members, there are limits on how much effort can be spent solely on administrative activities. Attaining the proper balance between production and control functions in complex systems is an important though sometimes ideologically sensitive issue.
b). Substitution: Some systems are protected against breakdown by designing components that are extremely simple and highly interchangeable. The use of highly standardized parts which can be readily adjusted to perform differing specialized tasks is one way of providing for substitutability of components. Another method is to reduce each specialized task to its most elemental operations so that new components can quickly assume the role of failed components.
An obvious example of a complex system with high potential for substitution is an automobile assembly line. While workers are very highly specialized, their tasks have been reduced to such an elementary level that most tasks can be learned by a new worker in a few minutes.
At least two important factors limit the extent to which substitution may be used to minimize breakdown in complex systems. The first is the limit to which a given task can be broken down into simpler subtasks. At some point, further subdivision and simplicity becomes impossible. The second limit is the extent to which workers are able to function effectively while performing extremely simple tasks repetitively. When the task becomes so meaningless that continued operation becomes difficult for even the most disciplined worker, efficiency falls off precipitously. Investigation and experimentation with these limits is currently underway in a number of industries throughout the world.
c) Redundancy: In some systems, the possibility of breakdown is reduced by deliberately providing extra copies of all critical components and interactions. These extras are either stored away in inactive status until needed or else all components normally operate at less than full capacity with each member of a redundant group capable of assuming a full load if one member should fail. Similarly, when it is vital that a particular interaction such as a message be delivered absolutely correctly, it is common to have the message transmitted several times, possibly through several different media, with the contents of each transmission compared to insure against error.
Redundancy is similar to substitution except that redundant components may be quite complex and still allow for replacement in the event of failure. Clearly this is an expensive method of preventing system breakdown since it increases the cost and size of the system being considered. Redundancy is usually only used where it is absolutely critical that a particular system perform properly. Thus redundancy is common in military and medical systems with spare parts and personnel kept in "storage" against the possibility that they will be needed in an emergency. Similarly, most space vehicles have totally redundant batteries, computers, fuel systems, and so on to ensure completion of the mission even when a critical component fails in space. Although most systems cannot afford to rely extensively on redundancy as a protective device, most automobiles carry a redundant "spare tire."
d). Self-regulation: Many systems contain mechanisms within themselves which are capable of repairing or correcting malfunctions within the system. While linked to administrative activities, such mechanisms are part of the internal operation of the system and operate automatically, usually without human intervention. Such self-regulating mechanisms are often referred to as "feed-back" or "cybernetic" systems.
The manner in which feedback systems help to regulate systems is fairly complex and will be discussed in greater detail in the next section. At this point it is enough to point out that the common thermostat is a well known negative feedback system maintaining the temperature of a room at a pre-determined level without direct human intervention.
e). Decentralization: A final method of protecting against system breakdown is to decrease the level of complexity of the system. This means decreasing or limiting the degree of specialization of components and increasing their self-sufficiency. Then, if a particular component or interaction fails, the remaining components will be minimally affected and the system as a whole will continue to operate. Examples of decentralization are found in many military units, each of which is designed to be capable of operating independently of other components which might be damaged or destroyed in battle. Similarly, many large corporations have been organized into a number of highly independent divisions in order to limit the need for centralized administration and sometimes to encourage competition among various divisions of the same corporation.
During the early 1950s some civil defense officials argued for decentralizing the United States in order to decrease the impact of losing several major cities through atomic attack. These strategies involved increasing local and regional self-sufficiency in order to decrease interdependence among various parts of the country. Similarly, the movement towards living in small, self-sufficient communes and rural farms prevalent during the 1960's has been viewed by some as preparation for the breakdown of our highly complex urban society. Clearly, decentralization is limited by the extent to which mutual dependence can be reduced while still maintaining a meaningful and productive system. The concept does, however, introduce the idea of deliberately limiting increases in specialization and connectivity in systems.
All five of these defense mechanisms are commonly employed in the design and maintenance of complex systems with no one mechanism necessarily superior to the others. Well designed and managed systems are likely to employ a combination of several of these mechanisms to protect against system breakdown.
Many systems contain some form of interaction between components which transmits information on system performance rather than actual energy, materials, or other products of conventional component functions. Such information "feeds back" to the producing component and tells it whether or not to modify its output to other parts of the system. The most commonly known form of such self-regulation is the negative feed-back system provided by a simple thermostat linked to a furnace.
The heart of a thermostat is a bimetallic strip which bends in one direction when warmed and in the opposite direction when cooled. When the temperature in the room falls below some pre-set minimum, an electrical contact is made and a small amount of electricity flows to a fuel control which increases the amount of fuel available to the furnace. With more fuel the furnace produces more heat and the room begins to warm up. As the room warms, the bimetallic strip in the thermostat bends in the opposite direction until the contact is broken and the flow of electricity to the furnace stops, decreasing the fuel supply. As the room slowly cools, the bimetallic strip begins once again to bend to the right and at the preset temperature contact is again made, increasing the fuel supply to the furnace and repeating the cycle.
Such a system is called a negative feedback system because an increase in one component (heat) results in a decrease in a second component (fuel). Negative feedback systems are self-limiting or homeostatic in operation, producing systems which are relatively stable within the constraints placed on them by their surroundings. That is, the furnace and thermostat system produces a stable room temperature only as long as fuel and electricity are available and external temperatures do not become excessive.
Several additional examples of negative feedback systems will help to elucidate their properties further. In a simple food chain consisting of foxes and rabbits, if the foxes eat too many rabbits their food supply is limited and foxes begin to starve. But as the number of foxes decreases through starvation more rabbits survive and the foxes' food supply begins to increase. A relatively stable equilibrium results within which a certain number of foxes can exist in the same habitat with a certain number of rabbits. The actual size of the two populations is determined by their breeding habits, the rate of growth of the rabbits' food supply, and by other factors external to the system under consideration such as weather and other predators.
A more prosaic example is seen in nursing babies. When a baby is not getting enough milk, hunger stimulates the sucking reaction and the baby nurses more vigorously. More strenuous nursing stimulates the mother's mammary glands and increases milk production. Increased milk production satisfies the baby's hunger more quickly, reducing the need for vigorous nursing. When the system operates properly the result is a satisfactory equilibrium with the mother producing just enough milk to satisfy the baby's needs.
The most widely known and sometimes misunderstood example of a negative feedback system is the so called "law of supply and demand." In a free market, when a given commodity is in high demand due to either shortages or high quality, the many persons seeking the commodity bid up its price, thereby increasing its profitability. Producers of the commodity then seek to increase their profits by producing more of it. Higher production satisfies more of the demand and eventually prices decline. As profits decrease fewer numbers of the commodity are produced and an equilibrium is finally reached where enough of the commodity is produced to meet consumer demand at a price low enough to attract consumers but high enough to yield a reasonable profit for its producers. This is Adam Smith's "Invisible Hand", seemingly controlling the production and distribution of goods without the need for production quotas, marketing orders and other paraphernalia of centralized planning and administration.
In such cases, the profit attained on a commodity serves as information provided to the producer regarding the buyers' appreciation of his product. This information is used to make production decisions about this and other products and with enough actors working fairly independently, a relatively well ordered self-regulating system results. The system does not operate properly when too few producers or consumers are involved or when the feedback information is distorted or misused. Ideological confusion occurs due to the fact that while money in the form of profits is used as information, it may also be used as individual wealth and capital. The fact that profits have begun to be allowed in certain markets in the Soviet Union is not a sign that they have embraced capitalism. Rather, it indicates that the Russians have begun to understand the significance of negative feedback systems in helping to regulate production and consumption.
So far only negative feedback systems have been discussed. Equally important though less well understood are positive feedback systems. Rather than the stability and equilibrium produced by negative feedback, positive feedback promotes change and instability in systems.
The negative feedback thermostat/furnace system can be converted to a positive feedback system by simply switching the wires on the thermostat so that if the room cools the contact breaks and the furnace is turned off. If the room warms instead, contact is made and electricity flows to the fuel control producing more heat. With the thermostat wired in this fashion, the initial presence of heat produces more heat which in turn produces more heat and so on up to the limits of the furnace and the fuel system. Given an unlimited fuel supply, a furnace of unlimited capacity, and an initial condition of heat in the room the system would eventually burn itself up. If the initial condition were a cold room, the furnace would never turn on and the system would eventually reach maximum coldness as determined by the outside temperature. The ultimate behavior of such a system is to maximize change in whatever direction it is initially started up to either the limits imposed by external factors or to the point of system self-destruction.
Obviously, being able to distinguish between positive and negative feedback systems is very important. A simple convention for identifying positive and negative flows within fairly complex systems and evaluating their total effect helps to identify positive and negative feedback systems.
Positive signs are assigned to flows involving direct relationships and negative signs to flows involving inverse relationships. After the signs are carefully designated, the total number of negative signs present is determined. If the total number of negative flows in the system is an odd number the overall result is an inverse relationship and the system as a whole constitutes a negative feedback system. Conversely, if the total number of negative flows is an even number, the inverse relationships have all cancelled each other out and the total configuration constitutes a positive feedback system.
Unfortunately, this method works only if all flows within the system are clearly identified and correctly labelled, a result which is quite difficult to attain even in systems of only moderate complexity. A more pragmatic way of distinguishing positive from negative feedback systems is to carefully observe their performance to determine whether they produce change or stability.
An important and very instructive example of a positive feedback system is found in the relationship between gasoline consumption and gasoline taxes used for highway construction. During the early part of this century operators of automobiles sought more paved roads upon which to drive their relatively fragile vehicles. When governing bodies were asked to provide such amenities, it was argued that those seeking and using paved highways should pay for them through taxes levied on some common component of their vehicles. Since most automobiles were by this time powered by gasoline, it was decided that if highway construction were paid for by a tax on gasoline, the costs of construction would be paid for only by automobile owners and in direct proportion to the amount they used the highways.
This solution seemed admirable at the time and did work quite well for many years. However, this established the positive feedback system illustrated in Figure 2.6 which operated continuously for over half a century. As more paved highways were built, automobile trips became easier and longer. As trips became easier and longer more people traveled further using automobiles and the number of miles traveled by automobile increased. As the number of miles traveled by automobiles increased, more gasoline was consumed. As more gasoline was consumed the amount of taxes collected on gasoline sales increased. As the amount of gasoline taxes increased, more highway construction was possible. As more paved highways are built, automobile trips become easier and longer. And thus the cycle repeats itself with all relationships direct and positive.
If no other constraints existed on such a system, the eventual product could be a nation made up of nothing but paved highways with the entire population riding around in automobiles. While we have recently managed to bring this system under some control, when the U.S. lifestyle in 1980 is compared to that of 1910 it is apparent that we have come a considerable distance towards paving much of the country and living in automobiles.
The system described by Figure 2.6 is a positive feedback loop which has contributed significantly to massive changes in our entire social system. The extent of these changes was neither intended nor anticipated when this method of funding highway construction was first proposed.
Another interesting attribute of positive feedback systems is that if started into a downward spiral by some external event, they can also run towards self-destruction rather than continuous growth. Thus, for example, higher gasoline prices in the United States during the late 1970's led to significant decreases in the number of miles being driven. This in turn reduced tax revenues used for highway maintenance resulting in poorer quality roads. As roads declined in quality, people became more reluctant to drive leading to less gasoline consumption, lower tax revenues and so on. This trend was finally recognized in the early 1980's and efforts were begun to protect the substantial investments already made in the national and state highway networks.
Numerous additional examples of positive feedback systems can be found. When the effects are undesired they are sometimes called "vicious circles." The "cycle of poverty and ignorance" is a common term referring to the tendency for poorly educated and unemployable parents to raise their children to be equally poorly educated and unemployable.
Negative feedback systems are sometimes converted to positive feedback systems. When nursing mothers worry whether their babies are receiving enough milk, crying or strenuous nursing is interpreted by the mother as a sign that her milk supply is inadequate. This leads to emotional tension which in turn inhibits milk production. Lower milk production leads to more crying and strenuous nursing and additional concern by the mother. The initial existence of this concern coupled with the ease and certainty of bottle feeding has cycled many modern mothers out of being able to breast feed their babies.
Positive feedback systems can produce change in either desirable or undesirable directions depending upon the initial status of the system. A small push in the proper direction to a positive feedback system causes it to continuously change itself in the desired direction without the need for further intervention. Being able to identify and properly start positive feedback loops can be an extremely useful technique for planners and other social change agents.
Positive feedback systems can also be modified to introduce some negative feedback, thereby bringing the changes being produced under some degree of control. An excellent example of this is again provided by the gasoline-highway construction system described earlier.
In recent years a number of states have attempted to use some of the money collected in gasoline taxes to improve publicly owned mass transit systems. Although strongly opposed by a number of special interest groups including automobile clubs and the highway construction industry, this reform measure has been passed in a number of states. The result is a negative subloop within the larger positive feedback loop described in Figure 2.6 which serves to slow down the rate of highway construction and growth of automobile usage. Figure 2.7 illustrates this more recent version of the gasoline-highway system.
In this case, as more gasoline taxes are collected, more money is available for mass transit improvements as well as highway construction. This makes travel by mass transit easier and attracts some formerautomobile riders to use the mass transit system. This then has a negative effect on the number of miles traveled by automobiles and introduces some negative feedback within the system. Fewer miles traveled by automobile means lower gasoline sales and less gasoline taxes collected. This is turn means a little less money for both highway construction and mass transit.
After several years of operation this revised system should produce a few less highways, somewhat better mass transit, and some proportion of the population which originally used automobiles for all travel now using mass transit for at least some of their travel. How large the change will be is determined primarily by the percentage of gasoline taxes allocated to mass transit improvements.
While property and income taxes are usually levied to raise money, some taxes are intended primarily to control the consumption of certain commodities like liquor and cigarettes. Both forms of taxes insert feedback loops into the system, however, even though regulating behavior was intended only in the latter case. If cigarette or liquor taxes are increased to raise more money, they may instead produce less revenue due to lowered consumption or increases in smuggling. Similarly, property taxes may also modify behavior. A tax on windows in medieval England was intended to collect more money from the wealthy who could afford windows. Another result, however, was to decrease the number of windows used in buildings. Real estate taxes based on building values tend to inhibit improvements and reward deterioration. Public officials must be aware of the feedback effects of taxes and distinguish carefully between those intended to provide revenue and those intended to regulate behavior.
5. Systems Analysis
It is useful to refer to the components of a system as "subsystems" and to the larger system surrounding the system being considered as the "suprasystem." If the component parts of a subsystem must be referred to they are then called "sub-subsystems."
In most cases, however, the identification of only three system levels is all that is required. The convention is to refer to the most important system on which attention is focussed as the "system" and to use the terms suprasystem and subsystem for the immediately higher and lower level systems involved. Despite this tri-level distinction, all systems at all levels are assumed to be similar in structure and functioning. More elegantly stated, all systems are assumed to be isomorphic.
Systems analysis consists of a six step process of determining how a particular system functions in terms of its subsystems and suprasystem. The same term is often used to refer to highly quantified investigations but elaborate measurement and high precision are not the critical ingredients of systems analysis. While seemingly quite simple, if carried out rigorously systems analysis can be an extremely powerful analytical tool which has the additional virtue of providing a clearly defined point at which the analysis is completed. The required steps are as follows:
a). Identify the system being studied. This is the most critical step and it is here that most errors are made. Many analysts are not really sure what they are studying and answer this question with popular but poorly defined terms like "voting behavior", "taxation", "health", and so on. Without more specification, it is hard to imagine what systems these terms refer to and it is very difficult to identify their component parts in system terms. Preferably, the studies referred to above would be defined as studies of the national or state election system, municipal or national taxation, and either hospitals, the medical profession, or perhaps the human body.
A related problem is focussing on the wrong level of system. Thus when studying runaway children one might study patterns of interaction among runaways, or the families of runaways, or even individual children. It is obvious that very different interests are represented by each of these choices of systems and very different findings and recommendations will result from each of these different investigations. It may be most appropriate to study all three of the systems indicated in order to more fully understand runaway children but this requires three separate analyses each with differently defined systems, subsystems, and suprasystems.
Once the system to be studied has been identified, its closure, connectivity, and stability must be assessed. While precise measurement of these characteristics is unlikely, careful examination of the system should provide a reasonable basis for comparing these properties to those of other systems of similar type. If any of these three properties appear to be unusually low or high, it is possible that the particular system under analysis is abnormal. It may be appropriate to redefine the system being considered or to select another example for analysis. If the analysis of the system originally selected continues, findings must be interpreted in the light of this apparent abnormality.
b). Identify the subsystems. Once the system has been clearly defined it should be broken down into its major components. This is a problem similar to that of adequately identifying the system in that each subsystem can itself be considered to be a system at a lower level of analysis. Often, the most appropriate subsystems are most readily identified by the existence of associated properties such as spacia[ separation of major subsystems, strong boundaries around subsystems, or other indications of relatively high closure, connectivity and stability. Identifying potential subsystems may also be aided by locating major functional activities of systems, possibly in terms of the "requisite functions" of systems which are discussed below.
When a tentative set of candidates for subsystems analysis has been identified, they in turn must be evaluated for closure, connectivity and stability in comparison to other subsystems of similar type. In general, subsystems are expected to exhibit lower closure and higher connectivity than the parent system within which they are embedded.
c). Identify the suprasystem. To the extent that the suprasystem is defined as everything outside the system, this is a relatively simple task. Some additional care must be taken, however, to isolate the major elements in the suprasystem with which the system most closely interacts. These elements need not be specified in any detail but a reasonable list of major external interaction partners is needed.
d). Determine the flows between the system and the supra system. This step is sometimes undertaken in great detail with highly refined measurements. While such precision may be useful, cruder assessments may be almost as valuable and require considerably less time and effort. What is needed is some coherent listing of the major products which flow from the system to the suprasystem and vice versa. Identifying the products and providing some indication of the frequency and volume of the interchanges is the most important aspect of this step. Precision and detail is of secondary importance.
In completing this step of the analysis it is important to include all kinds of interchanges, both tangible and intangible. It may also be necessary to consider interactions which have occurred or will occur in the past or in the future. Thus shipments of iron, coal, potatoes, and automobiles are fairly obvious products to be considered. A little less obvious are telephone calls, bank deposits, recreation, and information through mass media. Still less tangible are commodities like loyalty, affection, and moral support in times of crisis. Very difficult to account for are promises and obligations which exist for actions taken in the past or actions promised in the future. Among such cross-temporal connections, funds from a pension earned 20-50 years earlier are among the more easily represented.
It is obvious that for any moderately complex system a complete listing of all interactions is not possible. But good approximations can be fairly easily developed and usually embrace 20 to 100 items for a moderate sized city. Included in the list are usually four or five classes of manufactured goods; newspapers, television and radio; major types of foodstuffs; a number of services such as education, medical care, and entertainment; governmental activities including highways, police, fire, and taxes; direct organizational affiliations; and population flows. It merits repeating that a comprehensive list with rough indications of volume and frequency is far more valuable than a set of very precise measurements of a limited range of commodities.
e). Determine the flows among the subsystems. This is basically the same as the work done above except that now the types of products are considered to and from each of the subsystems. The same lists established for the total system flows and the same standards of measurement may be employed. The advantage of having identified a relatively small number of subsystems in the second step becomes apparent at this point.
f). Balance the flows among subsystems with the flows between the system and the suprasystem. Each of the subsystems consumes certain commodities and produces others. When the flows among all subsystems are summed up, there should result a net surplus of some commodities and a net deficiency of others. These net products represent the differences between production and consumption among all the subsystems taken together. In a balanced system, the net surpluses and deficiencies should be approximately equivalent to the net flows out of and into the system from the suprasystem. If these commodities are considerably out of balance it is reasonable to conclude that the system as a whole is in a state of change or that a substantial error has been made in one of the first five steps of the analysis.
If the surpluses and deficiencies balance reasonably closely with the flows between the system and the suprasystem, the systems analysis is completed. The investigator now has a good picture of how the system operates and which of its component parts produces and consumes the various commodities with which the system interacts with the outside world.
Treating subsystems as "blackboxes" in this fashion, i.e., noting only their inputs and outputs, does not provide an understanding of how the subsystems themselves operate. If it becomes important to understand how a particular subsystem operates in producing the interactions with other subsystems which make up the functioning of the system, one need only undertake a new systems analysis at the appropriate subsystem level. In this case the subsystem is designated as the system and the steps outlined above are repeated. Additional analyses at successively lower system levels can be repeated until even the most demanding critic is satisfied with the level of understanding provided.
6. Levels of Order
Many observers have noted the existence of several different kinds and levels of organization in systems, particularly in human social systems. Thus all systems exist within some physical and environmental context which sets some limitations and determines certain properties of the system. Sunlight, temperature, water, soils, and so on are all obvious physical determinants of the kind of living systems which may exist in a locality. They provide the PHYSICAL ORDER within which the living or BIOTIC ORDER develops.
The biotic order together with the physical order in turn help to determine the kinds of raw materials, crops, and manufactured products to be found in an area. These properties then determine to a considerable extent the kind of livelihood and trade in which inhabitants of an area are able to engage, including trade with other areas possessing different physical and biotic properties. Trade and commerce, however, require agreements on the materials and terms of trade, monetary systems, and so on which evolve into an ECONOMIC ORDER.
Since stability and peace are desirable for efficient economic relations, governments, police forces, laws, and courts often emerge alongside flourishing economies. Thus a POLITICAL ORDER is seen by some to be based upon an underlying economic order. In addition to formal legal and governmental controls, social standards for behavior also develop creating an additional form of order and regularity of behavior which may be called the SOCIAL ORDER. Finally, the belief and value systems of the people create a moral or IDEOLOGICAL ORDER over and above their day to day practices. This order reflects the things people believe in and what provides their motivations, regardless of their individual or group behaviors.
Among others, Robert Park has described these various levels of order in human systems and has suggested they form a pyramid with biotic order at the base and cultural or moral order at the apex; with economic and political orders intermediate. While a useful description, Park's argument and illustration has suggested to some that understanding of the orders at the base is fundamental to understanding human society while political and ideological issues are epiphenomena derived from these more basic factors. As our discussion of feedback systems has already shown, however, making deterministic arguments among elements of a system is both simplistic and misleading. No single component determines the properties or behavior of others. All components effect all others and explanations or theories based on single factors should be viewed as simplistic representations of only a part of the interlinkages among system components.
Rather than a pyramid, the relationship among levels of order is better represented by a shape like that in Figure 2.8, more a vase than an inverted pyramid. Here the various levels of order are represented in a manner similar to Park's pyramid but the increasing width of the figure reflects the increasing variety to be found in the various types of order at differing levels. This illustration should reduce the acceptance of deterministic arguments since it is readily apparent that many different types of economic, political, social, and ideological orders can emerge from an underlying physical and biotic order. Understanding the limitations imposed by the physical and biotic orders is necessary but never sufficient to understanding political or cultural orders.
It should now be obvious that understanding human systems requires that all the types of order at all levels be given full consideration. Thus an interpretation which reflects only biological, or economic, or psychological factors is inadequate by definition. Full understanding of system behavior requires a multidisciplinary approach capable of understanding the ordering of the system at all levels and the interrelationships among them. In the chapters which follow interpretations and theories for each system considered will be broken up into three chapters, one on physical properties reflecting largely physical and biotic explanations, one on institutional properties reflecting primarily economic and political explanations, and one on behavioral properties reflecting primarily social and cultural explanations.
7. Requisite Functions
A number of theorists have found it fruitful to try to specify the requisite functions necessary for a system to operate. Each of these functions must be performed by one or more of its subsystems or the goods or services provided by the function must be imported from some outside the system, presumably in exchange for the surplus products of some other requisite function performed within the system. Any system which fails to provide for the fulfillment or acquisition of all requisite functions must either change or die. Although all requisite functions are necessary for system survival, some may be more critical than others.
The number of requisite functions specified by various theorists ranges from three to fifty depending upon the needs and interests of the particular theorist. In some cases functions are defined in terms of individual needs while in others they are formulated as organizational requirements. A small selection of some of these interpretations of requisite functions is provided in the bibliography at the end of this chapter. For our immediate purposes, however, the important point is to understand the concept and utility of requisite functions.
Employing any set of reasonably well defined requisite functions, a system may be examined for the extent to which it meets its own requirements in each of these areas. Deficiencies suggest either dependence on other outside systems or an imbalance in current system operation which will lead to change. A surplus suggests either a functional activity in which the system specializes and trades with other components in its suprasystem or a system imbalance. Identification of subsystems is also aided by attempting to locate the components of systems which fulfill each of the critical functions. Additionally, interactions among system components and between system and suprasystem may be categorized according to the requisite functions involved.
Among the many theorists defining requisite functions, Herbert Spencer (1897) identified three critical activities: Production, Distribution, and Control. To these, Amos Hawley (1957) added a fourth: Recruitment and Training. Based on the work of Durkheim and Weber in addition to more recent group process observations, Talcott Parsons (1965) identified a somewhat similar set of four functions: Adaptation, Gratification, Integration, and Latency. The U.S. Army and many large corporations also recognize four major types of subunit activity resembling these four functions.
While employing a minimum number of categories, Hawley's four functions appear to provide a minimal comprehensive classification of all the important activities occurring in both simple and complex societies. We will use these four functions to describe the ways in which social systems provide for the fulfillment of functional requisites. To distinguish them from the variety of terms used by other theorists and organizations, they are here termed: Production, Allocation, Control, and Staffing and are also referred to as the "PACS functions."
Production functions are activities providing adaptation to the environment, converting raw materials into usable products. In its most specific interpretation production includes gathering, hunting, fishing, farming, mining, forestry, and manufacturing. More generally, however, production functions include adaptation to the suprasystem as part of the external environment. Under this broader interpretation the production of any any surplus commodities or services for "trade" with the suprasystem is considered a production functions.
Mines, factories, and farms are the usual production components of systems but in a typical college town the college itself must be recognized as a production activity. The same is true of recreational activities in a resort city such as Las Vegas; of medical services in a city like Rochester, MN - home of the Mayo Clinic; or of governmental services in local, state, and national capitals.
Allocation functions provide members of the system with the goods, services, and information they need. Activities normally included under allocation are the buying and selling of goods through retail and wholesale establishments; transportation activities; and information dissemination. In some instances it is appropriate to also consider the distribution of non-material rewards and punishments as part of the distribution function. Many cities specialize in providing such functions for their surrounding regions.
Control functions are those activities devoted to administering and defending the operation of the system including the creation and enforcement of laws and defending the system from external threats. In cities, such activities typically include most aspects of city government, especially the city council, police force, courts, and certain regulatory offices. In business organizations, control functions involve many but not necessarily all management functions down to supervisors and foremen. County, state, and national capital cities and those grown up around military bases are centers of control functions, providing these activities for the larger systems of which they are a part.
Staffing includes a very broad set of activities which are often overlooked as necessary functions within systems. Included as part of staffing are the recruitment and training of new members of the system through births, migration, recruitment programs, and schools. Also included are a variety of personnel maintenance functions such as medical, dental, and psychiatric care as well as activities providing rest and recreation for members of the system. Subsystems providing staffing functions include families; schools and colleges; hospitals; medical, dental, and psychiatric clinics; parks and recreation areas; churches; and various entertainment and recreation centers. A number of cities can be identified which specialize in providing staffing functions for the larger system within which they are embedded.
The requisite functions of a system can be an exceptionally useful tool in analyzing and interpreting systems. Some caution is called for in asking too much of this relatively simple concept, however. When analyzed closely, most system activities and components seldom fit conveniently under such neat and convenient theoretical labels. Single subsystems often fulfill several different functions while a single function may be provided by several different subsystems.
8. System Properties
A system at any level is a functioning organism in its own right and has a number of properties which belong uniquely to the system but not to its component parts. Such emergent properties are call system properties. A few examples for cities will help to explain this concept and show how system properties differ from the properties of members or components of the system.
The age of a city refers to the number of years since its founding, although sometimes it is measured by the number of years since it reached some important stage of maturation such as becoming a capital or attaining a population of 50,000. A city's age is usually unrelated to the age of its citizens, however. The median age of its citizens may be 27.6 years, while the city itself may anything from a few to several hundred years old.
The size of a city is usually measured in terms of either its total population or the number of square miles it covers. City size has nothing to do with how heavy or tall its citizens may be and is not necessarily related to the size of its buildings. Similarly, cities may be said to have a certain type of government, a certain type of industrial specialization, a particular type of spatial pattern, or ahigh crime rate. Most of these characteristics are also unrelated to individual properties of citizens or other components of the city.
The problems sketched here are commonplace but are frequently overlooked. Fortunately, aggregate statistics and the inferences drawn from them often do approximate system properties. The potential for faulty interpretation is found everywhere, however, and even experienced analysts can be led astray when they fail to identify clearly the systems being studied and the distinction between individual, aggregate, and system properties.
A little less clear, however, are attributes such as median age, average wealth, or level of unemployment. These are also system properties belonging to the city but in this case the properties are summary characteristics of the properties of individual citizens. Thus system properties may in some cases reflect aggregate measurements of the attributes of system components.
Being able to distinguish between properties of systems and those which are merely the aggregated properties of system components is a very important theoretical and methodological skill which is closely connected to understanding clearly which system at which level is currently being investigated. Failure to clearly identify which system is being studied and to distinguish between system properties and the properties of members of the system can lead to dangerous misunderstandings and misinterpretations of data and research findings. The "ecological fallacy" is only one of a number of such errors which can occur through not understanding the significance of system propertiesand clearly defining the system under investigation.
9. The Ultimate System
One additional attribute of systems must be considered, although this topic creates more anxiety than clarification about systems and their properties. Most systems maintain themselves over time, adapting to changes in the environment and to internal pressures in order to continue operating as a system. While providing stability and continuity, this tendency also leads to considering the well being of the system as more important than the well being of its individual components.
Since all systems at all levels are presumed to obey the same principles there is no theoretically valid way to determine which, if any, system level should be granted ascendancy in terms of asserting its well being ahead of that of its subsystems or suprasystems. Given no basis for choice, the tendency is to grant higher priorities to larger systems than to smaller ones. But in this direction lies fascism with various acts allowed "for the common good," "for the good of the state," or even "in the interests of national security."
Although there are no criteria for determining which system at which level is most important, one particular type of system is quite distinctive and deserves special consideration and priority in analysis and policy. That system is the individual human being. Certainly it is true that most individuals are willing to sacrifice one of their component parts in order to save the rest of themselves - cutting off an arm, a leg, or removing various organs as required. Conversely, the activities of some organizations are sometimes modified when it becomes apparent that the rights and opportunities of members of the organization will be harmed unless the modifications are made.
Obviously, public decisions cannot always be made only with respect to the rights of individuals. In any civilized society it is imperative that the general public welfare may take precedence over the rights of individuals. However, when a vaguely defined and relatively trivial public benefit is to be attained at considerable cost to a few individuals, one should pause and reconsider the relative importance of individuals and the larger system and ask which one is supposed to be serving the needs of the other. Further discussion will not clarify or help in resolving these dilemmas. This brief comment, may, however, make more difficult some of the simple quick decisions sometimes made in the name of the larger systems. Policy makers should be willing to at least suffer a few sleepless nights when making such decisions "for the good of the system."
BIBLIOGRAPHY Chapter 2
The reader by Walter Buckley cited below contains reprints of over fifty articles on general systems theory including those by Ashby, von Bertalanffy, Boulding (1956), Churchman and Ackoff, Deutsch, Hall and Fagen, Hardin, and Rapoport and Horvath which are listed separately below.
Comments on the distinctions between simple and complex systems are offered in the readings by Alexander and Forrester.
Interesting discussions on the relationship between size and structure of systems are provided in the readings by Anderson and Warkov, Boulding (1956), Haire, and Terrien and Mills.
The articles by Aberle, Malinowski, Maslow, Miller, and Parsons and Smelser provide a variety of interpretations of the requisite functions of systems.
Aberle, David F. et. al "The Functional Prerequisites of a Society, Ethics, LX, No. 2 (Jan. 1950): 100-111.
Alexander, Christopher, "A City is Not a Tree, Architectural Forum, (Whitney Publications, 1965).
Anderson, T. and S. Warkov. "Organizational Size and Functional Complexity: Administration in Hospitals," American Sociological Review, Vol. 26 (February, 1961): 23-28
Ashby, W. Ross, "Regulation and Control," from An Introduction to Cybernetics (London, Chapman and Hall, 1956) pp. 208-218
von Bertalanffy, Ludwig. "General System Theory - A Critical Review," General Systems, Vol. 7 (1962): pp.1-20.
Blalock, H.M. and Ann B. Blalock. "Towards a Clarification of Systems Analysis in the Social Sciences," Philosophy of Science, Vol 26, No. 2 (April 1959): 84-92.
Boulding, Kenneth. "Toward a General Theory of Growth," Canadian Journal of Economics and Political Science, Vol. 19, No. 3 (August, 1953): 326-340.
Boulding, Kenneth, "General Systems Theory - The Skeleton of Science," Management Science, 2 (1956): 197-208.
Buckley, Walter. Modern Systems Research for the Behavioral Scientist: A Sourcebook. (Chicago: Aldine, 1968)
Churchman, C. W. and Ackoff, R. L. "Purposive Behavior and Cybernetics" Social Forces, Vol. 29 No. 1 (October, 1950): 32-39
Deutsch, Karl, "Towards a Cybernetic Model of Man and Society.," from "Some Notes on Research on the Role of Models in Natural and Social Sciences" Synthesis, Vol. 7 (1948-1949): 506-533
Feibleman, James K. and Julius W. Friend, "The Structure and Function of Organization," in Philosophical Review, vol. 54 (1945): 19-44.
Forrester, Jay W. "Counterintuitive Behavior of Social Systems," in Technology Review, 73, No. 3 (Jan., 1971).
Haire, Mason. "Biological Models and Empirical Growth Histories," in Modern Organization Theory, (New York: Wiley Press, 1959): 272-305.
Hall, A. D. and Fagen, R. E. "Definition of System," General Systems, 1 (1956): 18-28.
Hardin, Garrett. "The Cybernetics of Competition: A Biologists View of Society," Perspectives in Biology and Medicine, Vol. VII (Autumn, 1963): 61-84.
Hatt, Paul. "The Concept of Natural Area," American Sociological Review XI (August, 1946): 423-27.
Malinowski, Bronislaw. "The Group and the Individual in Functional Analysis," American Journal of Sociology, Vol 44 (May, 1939): 938-964.
Malinowski, Bronislaw. Argonauts of the Western Pacific, (New York, NY: E.P. Dutton & Co., 1961)
Maruyama, Magorah, "The Second Cybernetics: Deviation Amplifying Mutual Causal Processes," American Scientist, Vol. 51 (1963): 164-179.
Maslow, Abraham. Motivation and Personality. (NYC: Harper & Row, 1970).
McLoughlin, J. Brian. Urban & Regional Planning: A Systems Apoproach. (NYC: Praeger, 1969)
Miller, James G. "Living Systems: Basic Concepts," Behavioral Science, Vol 10, No. 3 (July 1965): 193-237.
Napier, Augustus and Carl Whitaker, "The concept of the System," Chapt 4 in The Family Crucible, (New York, NY: Bantam Books, 1980): 44-58.
Parsons, Talcott and Neil Smelser, Economy and Society, (New York, N.Y.: The Free Press, 1965): 13-28.
Rapoport, Anatol and Horvath William J., "Thoughts on Organization Theory," General Systems, Vol. 4 (1959): 87-91
Rapoport, Anatol, "Forward", in Buckley, Modern Systems Research for the Behavioral Scientist, (Chicago: Aldine Publishing Company, 1968): xiii-xxii.
Robinson, W.S., "Ecological Correlation and the Behavior of Individuals," American Sociological Review, XV (June, 1950): 351-57.
Spencer, Herbert. The Principles of Sociology, (New York, NY: Appleton-Century-Crofts, 1897)
Terrien & Mills "The Effects of Changing Size upon the Internal Structure of Organizations," American Sociological Review, Vol. 20 (1955): 11-13.