************

Working Paper: This document is made available for private study and comment in
anticipation of possible future publication, at which point the paper in its present form will be
removed from this website. You may print one copy for personal use, but do not
redistribute without written permission of the author.
*********

CHAOS AND THE NATURE OF ARCHIVAL SYSTEMS

William J. Maher
University of Illinois at Urbana-Champaign

Presented at the Society of American Archivists

56th Annual Meeting Montreal, Quebec, Canada
September 15, 1992

A new approach to scientific work and thought, termed chaos theory has recently

drawn considerable attention among researchers. This so-called "new science of chaos"

attempts to discover underlying structures in complex phenomena previously seen as highly

disordered and beyond deterministic control. Classic areas for chaos research involve

climate, fluid dynamics, and biological populations. 1 The 1970s' and 1980s' advances in

ways to find order or system within what was previously seen as impenetrable chaos is an

important moment in our ability to understand the universe.

To the non-scientist, these discoveries not only explain important natural processes

and the limits of deterministic mechanical systems, but also suggest a framework for

understanding human processes, such as documentation and archival practice. Just as the

past discoveries of Newtonian mechanics or Darwinian evolution have prompted us to adopt

1
Throughout this paper, the term "chaos" will be used to refer to the new perspective on
complex systems, and the traditional idea of chaos as disorder will be referred to as
randomness, disorder, or "utter chaos". For convenience, chaos will be referred to as a
science and as a theory, but it might be more proper to see chaos as a scientific vision of
diverse components of the universe behave. Thus, while "chaos" is treated rather uniformly,
in fact, the scientific research findings that have been the basis for the "new science of chaos"
might be better seen as a phase of development in the respective disciplines rather than as
part of a single overarching effort to understand chaos. Nevertheless, the fact that interest in
chaos has spread so rapidly in the 1980s suggests that its ability to explain phenomena in
diverse disciplines may lend it a transcendent value.
Chaos and the Nature of Archival Systems 2
William J. Maher 15 September 1992

the language and conceptual frameworks of scientific ideas, so too might the concepts of

chaos provide a rich means to understand and explain phenomena that hitherto have been

difficult to understand.

This paper will focus on the application of chaos to archival theory and practice. As

background it will describe basic elements in chaos theory and provide examples of chaos

research. A key part of the paper will be an examination of writings in archival theory for

evidence of ideas consistent with chaos theory.

THE SCIENCE OF CHAOS

In its simplest terms, the new view of chaos consists of discovering structure and

system within complex and turbulent phenomena previously seen as either lacking any order

or of such enormous complexity that controlling them or predicting outcomes would be

impossible. Nature is filled with systems that are either complex from their initial state, such

as the atmosphere, or are simple in their initial state but can go through a transition into a

turbulent state where predictability and control are impossible, such as a stream of cigarette

smoke that starts in an understandable stream but shifts into a swirling cloud. 2

Scientists have long been aware of complex systems whose behavior is dependent on

2
The three best brief introductions to chaos are: the introduction to N. Katherine Hayles,
Chaos Bound: Orderly Disorder in Contemporary Literature and Science (Ithaca, N.Y.:
Cornell University Press, 1990); Gary Taubes, "The Mathematics of Chaos," Discover 5
(September 1984): 30-39; and James P. Crutchfield, J. Doyne Farmer, Norman H. Packard,
and Robert S. Shaw, "Chaos," Scientific American 255 (December, 1986): 46-57. A more
comprehensive and well-illustrated discussion is provided by James Gleick, Chaos: Making
a New Science, (New York: Penguin Books, 1987), but Gleick overlooks important subjects
such as Russian and Japanese mathematical research and the work of Ilya Prigogine. Heinz
Pagels, The Dreams of Reason, (New York: Simon and Schuster, 1988) also provides a good
description of chaos and places it in the context of the history of science since the late
Renaissance.
Chaos and the Nature of Archival Systems 3
William J. Maher 15 September 1992

large numbers of variables. The traditional approach has been to attempt to understand such

problems by reducing them to their smallest parts in the belief that predictability could be

achieved by collecting enough data on each small component before attempting to

understand the whole. This has led to a tendency to ignore turbulent phenomena as being

essentially unsolvable and irrelevant to scientific work.

By contrast in the past few decades, several scientists have discovered that not all

turbulent or disordered phenomena are without structure. Instead, many behave in ways that

are scientifically describable. To categorize such phenomena, the term "chaos" has been

used in a new way to connote complex and unpredictable behavior which, while not

following simple mechanical laws, follows complex patterns that are mathematically

describable.3

There is not time today to do justice to the full story of "chaos." Rather, as a basis for

archival use of the language and ideas of chaos, I will describe a few concepts and cite

examples of chaos research. Probably, the most important concepts or attributes are:

sensitive dependence on initial conditions, strange attractors, feedback loops, fractals, and

recursiveness and self-similarity, and information richness. 4

3
Some scientists doing research in these fields have shunned the word chaos and instead
have preferred more respectable sounding names such as complex systems, non-linear
dynamics, or nonequilibrium statistical physics. Certainly these alternative names offer a
richer description than "chaos," and eventually the scientific terminology may evolve into
more precise and differentiated terms. For the moment, however, "chaos" has been widely
accepted in scientific and popular literature. In addition, use of the term "chaos" can
emphasize the revolutionary nature of many of these new scientific developments which are
explaining parts of the universe which were previously considered ununderstandable. Thus,
"chaos" will be used throughout this paper to cover all these complex phenomena.
4
Another key idea in chaos theory is that complex systems can create structure as they
dissipate energy. This may provide a clue to why applications of the idea of entropy cannot
Chaos and the Nature of Archival Systems 4
William J. Maher 15 September 1992

SENSITIVE DEPENDENCE ON INITIAL CONDITIONS

This key attribute of chaotic systems simply means that very small differences can

have big consequences and is sometimes called the "butterfly effect." It is often explained by

reference to Edward Lorenz's 1960s work on long-term weather forecasting. Using a

computer model of the atmosphere, Lorenz repeated a series of 13 equations several times,

using the output of one set as a source of data for the next run. The model worked

reasonably well until Lorenz stopped and then restarted the calculations using a printout

listing of the last value from the stopped run for the restarted run. However, the new model

began to diverge radically. (slide) Upon investigation, he discovered that while the computer

used numbers rounded to the sixth decimal place, the restarted run had used a number

rounded to the third place. Despite this very minor difference, the system behaved quite

differently. Based on such minor initial differences causing dramatically different

conditions, he concluded that long-range weather forecasting would be unlikely because such

a minor event as a butterfly flapping its wings over China could affect the weather on the

Eastern coast of the United States.

STRANGE ATTRACTORS

If chaos theory is pursued no further than sensitive dependence, we are left with the

conclusion that much of the universe is unstructured and unmanageable. The companion

concept of strange attractors, however, provides the structure. Strange attractors are the

explain increasing complexity in the universe, especially in evolution. According to entropy,

lifeforms should be winding down into ever simpler forms. However, if chaos processes are
part of evolution (e.g., populations dynamics), it may explain how the dissipation of energy
can lead to complex structures. Ilya Prigogine and Isabelle Stengers, Order out of Chaos:
Man's New Dialogue with Nature (New York: Bantam Books, 1984).
Chaos and the Nature of Archival Systems 5
William J. Maher 15 September 1992

boundaries of behavior of chaotic processes--they are the tendency toward order within a

system and they illustrate the range of conditions which normally occur within a complex

system even though one cannot precisely predict its behavior.

Again, Lorenz's work provides a good explanation. After his initial discovery, he

began calculating as many solutions to the equations as possible based on different starting

points. Rather than finding randomness, Lorenz found that the numbers varied only within a

certain range, and he plotted the data to reveal a visual image of an attractor which has

subsequently been called the "Lorenz attractor." As the following slides shows, the attractor

demonstrates a complex pattern with trajectories that never exactly repeat each other, but all

of which occur within a limit space or set of possibilities. In practical terms, the attractor

explains that July blizzards in Miami are outside of systems' boundaries.

FEEDBACK LOOPS

One of the reasons that small differences can have big consequences systems is the

succession of events where the output of one event becomes input for the next.

Consequently, minor differences at the outset, or those introduced by environmental

conditions, have increasingly dramatic effects as the events repeat and multiply. This can be

visualized as a taffy-pulling machine. Two points may be infinitely close on the taffy as it is

placed on the machine. With each successive cycle of the machine, the taffy is stretched and

kneaded, and the points move progressively farther away from each other. Thus, after only a

modest number of cycles, the points can be literally miles apart, and it would be impossible

to predict where they would be. 5

5
The taffy-pulling machine is an example of the horseshoe shaped mathematical model
developed by Stephen Smale. (slide)
Chaos and the Nature of Archival Systems 6
William J. Maher 15 September 1992

FRACTALS, RECURSIVENESS, AND SELF-SIMILARITY

The borders and patterns in complex systems are highly irregular, but still systematic

and possible to model through fractal geometry, a term and concept coined by Benoit

Mandelbrot. Fractals explain the additional fractional dimension found in the surfaces of

complex natural forms, such as coastlines or snowflakes, as well as strange attractors. This

can be seen in his development of a Koch curve (use slide). Because of the recursive nature

of the mathematics of complex systems, fractals illustrate the basic structure of a system such

that whether one looks at it from a distance or up-close, its nature is the same. The following

slides of Mandelbrot fractals illustrate this visually. If you have any doubts about the

relevance of this geometry to complex natural phenomena, examine the ice-crystals that form

on the inside of your windows on the next sub-zero day. (slide)

CHAOTIC SYSTEMS ARE INFORMATION RICH

Rather than seeing chaos as meaninglessness and an information "sink," the new

approach argues that these systems are rich information sources, provided we have the

equations or other descriptive tools to examine them. That is, the stretching and folding of a

system moving into chaos creates new information as it replaces the information used to

describe its initial state with that needed to describe each of its subsequent states. Thus,

rather than seeing the chaotic system as an example of entropy where the system loses its

information value as it winds down, Robert Shaw has noted that chaotic systems are be

information rich.6

6
Robert S. Shaw, "Strange Attractors, Chaotic Behavior, and Information Flow,"
Zeitschrift für Naturforschung 36a (1981): 80-110. See also, Crutchfield, "Chaos," pp. 53-
56.
Chaos and the Nature of Archival Systems 7
William J. Maher 15 September 1992

PHYSICAL AND SOCIAL SCIENCE APPLICATIONS OF CHAOS

The development of these concepts has significantly altered how we perceive and

describe complex natural phenomena in three ways: 1) orderly structures can be found

within disorderly phenomena, 2) such structures will be understood best by examining whole

systems rather than by reducing them into separate components, and 3) precise predictability

is no longer a pre-eminent test of whether we understand complex systems. To understand

the practical implications of chaos one need only look at some of the applications in the

physical sciences. Chaotic period doubling has been found in muscle fiber contraction in the

human heart just before the onset of an often fatal condition known as ventricular fibrillation,

and thus provides a way to understand a medical problem that had previously appeared to be

random and indescribable.7 Chaos is not always associated with negative conditions. Brain

cells may generate chaotic waves of electricity that thereby keep the brain in an active and

receptive mode to be able to receive signals and interpret the world. 8 Chaos has found in a

large network of connected computers which were designed to "bid" against each other for

access to central processing time. The slightest disturbance in the system, such as delays in

sending a string of data, could lead to highly unpredictable variations in the solutions that the

computers were calculating.9 Chaotic irregularities have been found in the patterns of a

7
Taubes, "Mathematics of Chaos," p. 34. See also Gleick, Chaos. pp. 283-85, 289-91; and
Briggs and Peat, Turbulent Mirror, 64.
8
Bruce Bower, "Chaotic Connections," Science News 133 (January 23, 1988): 58-59.
9
John Markoff, "In Computer Behavior, Elements of Chaos," New York Times,
September 11, 1988, p. ??
Chaos and the Nature of Archival Systems 8
William J. Maher 15 September 1992

sound signal even when other measures, such as the pitch, do not show noticeable changes. 10

As a bridge to understanding the role of chaos in archives, we can examine social

sciences and business management literature which has been quick to examine chaos to see

what relevance it can hold for these fields. A few examples illustrate the potential uses of

chaos theory. After looking at Wells Fargo's unforseen 1986 acquisition of Crocker National

Bank, Robert Waterman concluded management processes should be adaptive and

comprehend unpredictability because traditional mechanical and linear models of business

organizations do not fit their behavior so well as chaotic systems. 11

Ikujiro Nonaka examined several Japanese corporations to assess how organizations

can increase their intake of information. 12 Drawing on examples of the renewal of

corporations like Honda and NEC, Nonaka argued that more information of greater benefit to

the organization can develop when chaotic or turbulent conditions exist, and he

recommended that managers therefore look for ways to accentuate turbulence to create such

creativity.

Educational administration researchers recently have suggested that chaos can provide

a framework to explain crises confronting school officials. Griffiths, Hart, and Blair

examined a conflict in a southwestern school district and found "evidence" of sensitive

10
"Sign a Song of Chaos," Science News 133 (May 7, 1988): 300.
11
Robert H. Waterman, "Strategy in a More Volatile World," Fortune 116 (December 21,
1987): 181-82.
12
Ikujiro Nonaka, "Creating Organizational Order out of Chaos: Self-Renewal in Japanese
Firms," California Management Review 30 (Spring 1988): 57-74.
Chaos and the Nature of Archival Systems 9
William J. Maher 15 September 1992

dependence, dissipative structures, strange attractors, and feedback mechanisms. 13

INFORMATION SCIENCE AND HUMANITIES APPLICATIONS OF CHAOS

More directly relevant to archival interest are a few attempts to apply chaos to

information science, history, and literature.

Albert N. Tabah's examination of nonlinear dynamics of literature growth is

relevant to archivists because it involves the dissemination of information.

Suggesting a similarity between studies of fluctuations in animal populations and of

fluctuations in literature citation, Tabah quantitatively analyzed the literature of

superconductivity from 1966 through 1990. He found that periods of sudden bursts in

activity could be understood better through chaotic models, such as attractors, rather

than traditional exponential and linear models. Although more investigation is

needed, Tabah's study is instructive to archivists considering issues such as the

relationship between volume of records and value and volume and use. 14

The application of chaos ideas to the humanities is beginning to be considered by a

small group of scholars. At present, I am not aware of any experimental or quantitative

humanities applications of chaos, but some of the proposals are interesting because of their

connections to history and texts.

Historians George A. Reisch and Donald McCloskey have emphasized that history

13
Daniel E. Griffiths, Ann Weaver Hart, Billie Goode Blair, "Still Another Approach to
Administration: Chaos Theory," Educational Administration Quarterly 27 (1991) 430-51.)
Ultimately, however, the argument is not very convincing because the elements of chaos are
described in only the vaguest of terms and areas where quantification might be employed
(e.g., population changes) are overlooked.
14
Albert N. Tabah, "Nonlinear Dynamics and the Growth of Literature," Information
Processing & Management 28 (1992) 61-73.
Chaos and the Nature of Archival Systems 10
William J. Maher 15 September 1992

may be chaotic in the extreme dependence on initial conditions found in historical events.

used chaos as a new way to discredit the idea of universal historical laws by arguing that

history basically follows nonlinear dynamics with a high degree of sensitive dependence and

extensive feedback loops. Neither Reisch nor McCloskey offered anything resembling

scientific proof and neither pursue the relationship between orderly events. Nevertheless

they present an effective argument for refuting covering-laws history and the importance of

narrative as the main vehicle of historical knowledge since only description can take into

account the many minor matters upon which big events depend. 15

Perhaps the most thorough and well-reasoned applications of chaos to the humanities

is Katherine Hayles' book-length explanation of the relation of chaos to recent developments

in literature and literary criticism, especially post-structuralism and deconstruction. She

argues that both chaos and the literary trends involve attempts to understand the variability of

meaning in systems or texts. Of archival relevance, she explains the relationship between

information theory and chaos, emphasizing that the information richness of chaotic systems. 16

ARCHIVES AND CHAOS

The these applications illustrate how broadly the interesting ideas of chaos can be

applied. But, in themselves, neither the ideas and nor applications justify a linkage with

archival theory and practice. Some of you may share Maynard Brichford's initial and

continuing skepticism of the relevance of chaos to archives based on a very pragmatic view

15
George A. Reisch, "Chaos, History, and Narrative," History and Theory 30 (1991), 1-
20. Donald N. McCloskey, "History, Differential Equations, and the Problem of Narration,"
History and Theory 30 (1991): 21-36. 15
16
N. Katherine Hayles, Chaos Bound: Orderly Disorder in Contemporary Literature and
Science (Ithaca, N.Y.: Cornell University Press, 1990).
Chaos and the Nature of Archival Systems 11
William J. Maher 15 September 1992

of the nature and circumstances of documentation. Others may share my first intuitive

reaction to the ideas of chaos--that they explain much about archives. Because neither the

enthusiast nor the skeptic should be content with leaving the chaos image at merely the

intuitive, we must consider the problem of proving or testing the linkage.

At first, I thought that one might demonstrate the connection though mathematical

analysis of key archival statistics, e.g., growth in documentation, level of use, or relationship

of use to external factors. However, we lack standardized statistics, and the meaning of even

the best numbers currently available is so variable that it is unlikely quantification would be

usable. As a result, I despaired of demonstrating that archival systems and the historical

record behave as chaotic systems, and for a while abandoned the issue. However, I could not

escape the fact that the more I applied archival principles to everyday problems in

arrangement, appraisal, and response to user inquiries, and the more I re-read archival theory,

the more the chaos model seemed to explain why some things worked and why others did

not. Therefore, I have adopted an alternate approach to investigating whether the principles

of chaos are useful to archives. I believe it is possible to explore the relevance of chaos ideas

by looking at how they can explain core developments in archival theory and practice. More

research is needed, but a review of a few archival writings and a few more recent

contributions illustrates how chaos may be a very useful language for understanding the

nature of documentation and archival processes.

One of the most important works summarizing the development of archival theory in

the 19th and early 20th century is Muller, Feith, and Fruin's Manual for the Arrangement and
Chaos and the Nature of Archival Systems 12
William J. Maher 15 September 1992

Description of Archives.17 Written in Dutch in 1898, revised in 1920 and widely translated,

the Manual did not appear in English until 1940. It laid out 100 rules for the practical

handling of archives and manuscripts, and in so doing explained key archival principles,

especially for arrangement and descriptive work. Permeating many of the rules is the basic

concept that archives are organic entities. At many occasions, especially in the first 14 rules

relating to arrangement, the idea of organic wholes, we might say systems, is the wedge use

used to separate archival practice from the approach of historians and librarians. For

example:

". . . an archival collection is an organic whole, a living organism, . . . Every archival

collection has, therefore, as it were, its own personality, its individuality, which the

archivist must become acquainted with before he can proceed to its arrangement."

(MFF, p. 20).

The use of "organic" language is no accident or side-effect of translation, but the

imagery recurs throughout the text and becomes the basis for further explanatory metaphors,

such as that of the "skeleton."18 The reference to the skeleton is then used to support the idea

that minor flaws in the organization of files need not undermine the overall concept of the

material's arrangement. As in complex systems where each detail need not be precisely

defined for the system to maintain its overall character, archival arrangement can stand

17
S. Muller, J. A. Feith, and R. Fruin, Manual for the Arrangement and Description of
Archives, translated from second edition by Arthur H. Leavitt, (New York: H. W. Wilson
Company, 1940).
18
Rule 20 includes the statement: "The comparison with the animal skeleton, more in
harmony with the definition that the archival collection is an organic whole, naturally
emphasizes the unchangableness of those lines; . . ." (Ibid. p. 69).
Chaos and the Nature of Archival Systems 13
William J. Maher 15 September 1992

despite specific imperfections. "If however, he [the paleontologist] wishes to form for

himself a picture of the animal whose bones he has joined together again, he follows very

closely the general structure of the body and the shape of the bones, but he takes no account

of the accidental circumstances, e.g., that one of the animal's paws had grown bent because of

a fracture or that one of the ribs is missing." 19

When they come to explain the importance of respect des fonds, the Dutch employ a

naturalistic language that is consistent with the suggestion that what archives document are

complex systems that follow many of the principles of chaos. For example, Rules 10 and 11

stress the idea of dismemberment and relate the need to keep deposits together to the idea of

maintaining wholeness and completeness of the historical record. (MFF, pp. 36-40).

As in many archival texts, Muller, Feith, and Fruin do not provide much theory for

description. Rule 38, however, has relevance to the notion that archival documentation

constitutes chaotic systems. It states that before description can focus on details, the archivist

must form a clear concept of the "dominant idea that presided over the formation of the

documentation."20 Despite the difficulty of capturing the "dominant idea" of a record series,

the Manual's emphasis on the series summary description as the primary descriptive tool also

make sense from the standpoint of chaotic and complex systems. In such systems, it is most

important to describe the boundaries and general nature of the system's flow, and it is

unproductive to describe the system in an incremental or reductionist fashion. Thus, item

level inventories might be useful for specialized access or control, but archival

19
Ibid. p. 71.
20
Ibid. pp. 101-102, see also Rule 41.
Chaos and the Nature of Archival Systems 14
William J. Maher 15 September 1992

documentation is too highly fractal to enable such descriptions to explain what the series

really is.21

Before turning to American writings, it is useful to consider a statement by the

influential English writer, Sir Hilary Jenkinson, to illustrates how consistent ideas of chaos

are with classical articulations of archival theory. Jenkinson saw the structure and

complexity of records as emerging from a natural growth process.

"They [archives] came together, and reached their final arrangement, by a natural

process: are a growth; almost as you might say, as much an organism as a tree or an

animal. They have consequently a structure, an articulation and a natural relationship

between parts, which are essential to their significance: . . . Archive quality only
22
survives unimpaired so long as this natural form and relationship are maintained."

In itself, the idea that archives are organic does not mean they fit the mold of chaos,

but the way this language has been used shows great consistency with chaos principles of

complexity, recursiveness, and non-reductionism. The central theme of archives as organic is

particularly evident in the work of Theodore Schellenberg. 23

As in chaotic systems where one cannot predict precise relationships in advance,

Schellenberg noted that classification of archives could only be done a posteriori not from an

a priori scheme. "The classes should be established as experience attests . . . not . . .

21
Ibid., p. 109.
22
Hilary Jenkinson, The English Archivist: A New Profession, inaugural lecture at
University College, London, October 14, 1947: London, 1948, as quoted in Schellenberg,
Modern Archives, p. 19.
23
T. R. Schellenberg, Modern Archives Principles and Techniques, (Chicago: University
of Chicago Press, 1956).
Chaos and the Nature of Archival Systems 15
William J. Maher 15 September 1992

arbitrarily . . . on the basis of speculation . . . " in advance. (p. 63) Repeatedly, he rejected

all efforts to arrange and categorize records by predetermined, abstract, or universal systems.

Rather the records themselves, like a strange attractor would set the order. (pp. 22, 114, and

188)

Schellenberg and others have long since told of the practical and theoretical reasons

for the failure of the original arrangement system at the Archives Nationale (p. 169). The

ideas of chaos, which run counter to Newtonian mechanics offer a further explanation. The

Archives' "methodical" scheme was built from a classification mentality consistent with the

Enlightenment's reductionist approach to science. However, we have subsequently learned

that as in some of the physical sciences, reducing information systems into ever smaller parts

cannot advance our understanding of the whole or of the relationships of the parts.

Complexity in the small parts is so great that one cannot reverse the process and reconstruct

the whole from the parts because of the variability of the parts.

The principles that ultimately emerged as workable, respect des fonds and the two

Prussian contributions of provenance and respect for original order, emphasize relationships

and wholes, rather than parts. Admittedly, some archivists have not been comfortable with

the archival emphasis on deriving order and system from holdings rather than from some

logical constructs. However, rather than be embarrassed by such fuzzy and seemingly

unscientific principles, we can look to chaos to explain why the archival a posteriori

approach may indeed be scientific.

Schellenberg is a leading exponent of modern archival appraisal. His explanation of

its principles remains both practically and theoretically sound even if subsequent archivists

have added important refinements. The soundness of his approach to appraisal can also be
Chaos and the Nature of Archival Systems 16
William J. Maher 15 September 1992

understood if one considers the historical record as chaotic. As with a chaotic system,

Schellenberg rejected a reductionist or item-level approach to appraisal and instead

emphasized the consideration of relationships and the whole body of documentation. 24

In addition, Schellenberg recognized the complexity and fuzziness of appraisal more

than recent writers credit him. As with complex natural systems where linear formulae fail to

explain the systems' behavior, he noted "Techniques cannot be devised that will reduce the

work of deciding upon values to a mechanical operation." (p. 94) He seemed to understand

that present and future generations might desire standards to regularize appraisal, but he

cautioned against too much precision, and emphasized that the organic nature of records

made general principles far more useful than precise standards. Decisions on retention were

linked to the circumstances of the records creation, the nature of the information or evidence

they contained, and the likely interest of users in the evidence or information. Each of these

elements in turn was so multifaceted, that charting a uniform course for decision-making was

unproductive. As in chaotic systems, charting decisions with precision is impossible given

the highly fractal, unpredictable, and fundamentally chaotic nature of documentation. (pp.

133-60)

A final observation from Schellenberg is relevant. As in natural systems, complex

behavior can be introduced by changing only one variable--increasing the volume of

documents. Just as the pattern of dripping water moves into chaos when the flow is

increased, so too Schellenberg emphasized that the nature and arrangement of records

became much more complex as their volume and the number of agencies generating records

24
Ibid., p. 21.
Chaos and the Nature of Archival Systems 17
William J. Maher 15 September 1992

increased, for example in the U.S. federal government. 25 In all such technologically-driven

changes, the added complexity has important implications for the nature of archival practice.

CHAOS AND THE REVISION OF ARCHIVAL THEORY

So far my re-explication of archival theory from the perspective of chaos has followed

a very traditionalist course. Critics might suggest that I have been overly attentive to

traditional archival writers, many of whose ideas have been called into question by more

recent developments, and that I have not paid sufficient attention to the emergence of new

ideas in the past decade. In this light, it is useful to look at a few recent theoretical

developments to identify how principles of chaos may explain not only the soundness of

traditional ideas but also the kernel of changing ideas.

For example, the record group concept and the related idea that classification systems

should reflect administrative hierarchy have long been fundamental to American archival

practice. Since these principles were derived from European ideas of provenance, one might

think that Max Evans' 1986 well-stated argument for replacing the record group concept with

authority control would undermine my use of the principle of provenance as justification for

the relevance of chaos to archives. 26 However, when you examine Evans's argument closely,

there is great consistency between his rationale for abandoning record groups in favor of

authority control and the suggestion that archives are chaotic systems. In fact, the great

strength of Evans' argument is that he is attempting to address the fundamental chaos or

25
Ibid., pp. 36, 183. He also cites the example of the introduction of paper in the later 14th
century which led to expansion of offices and records systems, including registry offices, p
66.
26
Max J. Evans "Authority Control: An Alternative to the Record Group Concept,"
American Archivist 49 (1986): 249-61.
Chaos and the Nature of Archival Systems 18
William J. Maher 15 September 1992

complexity of archival information. He states, "The disadvantages of the record group

concept can be overcome by shifting the center of the archival world view from one that is

flat and mono-hierarchical to one in which records and the record creating agencies exist in a

multi-dimensional conceptual space." (p. 255) He wrote "Archivists must cast off the model

that holds that records have only a single referent and create a system that recognizes instead

that they are created and maintained as part of complex bureaucratic networks." (p. 261)

Evans' emphasis on the importance of context is consistent with how one should focus

on boundaries and relationships in understanding a complex system rather than on structures

extrapolated from individual elements. Evans thereby develops an approach which is more

consistent with both the core of the principles of respect des fonds and provenance and with

the fundamental non-linear dynamical nature of archival documentation. 27

BOLES-YOUNG

Given that Julia Young could not attend this meeting, it is with reluctance that I now

move to an examination of the work that she and Frank Boles have done on the "black box"

of appraisal. At the outset, I should acknowledge their great contribution of exploring and

describing the detailed aspects of appraisal. In particular, their categorization of three areas

of decision making--value of information, costs of retention, and political and procedural

implications--elucidates much of what happens when we appraise. 28 At the same time, I

27
A nearly contemporaneous article by David Bearman and Richard Lytle came to a
similar conclusion for using authority control as an alternative to the record group concept.
David A. Bearman and Richard H. Lytle, "The Power of the Principle of Provenance,"
Archivaria 21 (Winter 1985-86): 14-27.
28
Frank Boles and Julia Marks Young, "Exploring the Black Box: The Appraisal of
University Administrative Records," American Archivist 48 (1985), 121-40.
Chaos and the Nature of Archival Systems 19
William J. Maher 15 September 1992

think that their unsuccessful attempt to understand appraisal by breaking the decision down

into more than 58 small parts provides further confirmation of the relevance of chaos theory

to archives.

Boles and Young were aware of the difficulties their model might face because of the

dynamic of the appraisal process. (p. 124). They are to be commended for their persistence

and thoroughness in following through with an exploration of the black box. With NHPRC

funds they conducted an experimental project to test the three-part model by asking archivists

from 14 institutions to assign weights to each of 43 elements as an appraisal decision was

reached.29 Underneath the project was the assumption that before "macro-level" solutions to

the problems with appraisal methodology could be solved, "micro-level" tools and analysis

would have to be developed. Through quantification they hoped to unlock the appraisal

decision process. However, archivists found that "the system did not work as a practical

selection tool Neither the final ratio nor the module scores accurately predicted absolute or

relative record value." (p. 79)

Their response was the traditional reductionist approach--further refinement and

greater archival familiarity with quantification would permit a refined model and even an

"expert system" to reach appraisal decisions.(pp. 81-82) As they confronted these problems,

Boles and Young saw the limits of what they were attempting in ways that I suggest are

consistent with the fact that chaos explains many aspects of archival work. For example,

they acknowledged how the dynamic nature of appraisal undermined their work: "The

methodology used here made the assumption that the relationships between the various

29
Frank Boles in association with Julia Marks Young, Archival Appraisal (New York:
Neal-Schuman, 1991).
Chaos and the Nature of Archival Systems 20
William J. Maher 15 September 1992

elements in selection could be expressed through simple additive procedures, augmented

where necessary by multiplication." However, the complex nature of the relationships

required a model that recognized how the decision process changed as each element in the

relationship changed. (pp. 82-83)

From the perspective of chaos, it becomes crystal clear that the effort could not

succeed because of the fundamentally non-linear dynamical nature of both documentation

and decision-making. Given the number of feedback loops involved in each, I think the

mathematics will remain against it, and we should not spend further time using quantification

to understand appraisal. (cf. Boles 1991, p. 89) As physicist James Crutchfield has said,

"The consequence of measuring with only finite precision is that the measurements are just

not good enough--chaos takes them and blows them up in your face." 30

As with complex natural phenomena, reducing a system into ever-smaller parts does

not provide a reliable way to predict the behavior of the overall system. Such exercise will

be problematic not just from the standpoint of knowing what numerical values to assign to

each element, but because even with highly precise measures, the systems of documentation

and decision-making exhibit key chaos characteristics (sensitive dependence and

recursiveness) that make prediction impossible. We need to understand something of the

components of the appraisal decision, but looking at the larger picture will be more

productive. The comments of chemist/physicist Yoishi Oono are relevant here: ". . . even if

you don't know the chemical details, you can know something. The chemical details are

interesting but you don't need them. It's like taking the computer down to quarks, but do you

30
Quoted in Gary Taubes, "The Mathematics of Chaos," Discover 5 (September 1984):
32).
Chaos and the Nature of Archival Systems 21
William J. Maher 15 September 1992

need to know that? There are structures independent of these details."

CONCLUSION

I am not so foolish to claim that chaos provides a universal theory of archives,

although so much of what we document as well as the information service we provide is

fundamentally chaotic. For now, I wish only to note that the concepts of non-linear

dynamical systems provide an additional significant explanation as to why some archival

ideas have worked and others have not. At the very least, as we refine and advance archival

theory, we must reckon with the principles of chaos or we are likely to develop overly

complex tools that ultimately will collapse of their own weight as they become ever more

artificially structured, ever less functional, and ever farther from the dynamical and chaotic

nature of human activity, documentation, and information needs.

Some may react to my suggestion of the relevance of chaos to archives with a "So

what! The phenomena you note are nothing new and have always been known." In

response, I suggest these phenomena may have been experienced and recognized, but before

the development of the ideas of chaos, we have lacked a language and a structure in which to

place them. Rather than merely dismissing the seemingly random or irregular behavior of

documentation, organizations, or user inquiries as unscientific or anomalous, we now have a

way to understand that these phenomena fit a pattern that has scientific parallels and may

corresponds to scientific laws. In the end, maybe I am advancing a theory of archives, at

least in the sense that theory may be seen as the product of nearly "self-evident" truths.

Finally, I started thinking about chaos with the intuition that it made sense-it seemed

to explain much that worked and did not work in archives. Subsequently, I have worried

about how to prove that archives are chaotic systems. For a while I set aside the idea as little
Chaos and the Nature of Archival Systems 22
William J. Maher 15 September 1992

more than an passing metaphor. I still believe it highly unlikely we can ever quantify the

world of archives to permit mathematical tests of the chaos model. Nevertheless, I am now

persuaded that the intuition should not be ignored. Chaos explains much about the ordered

disorder that archives attempt to control, and it helps us better understand the order parts of

the universe we confront. 1

I find myself taking heart from Fred Stielow's thoughts on the nature and usefulness

of theory when he quoted Lionel Robbins: "'We do not need controlled experiments to

establish their validity: They arise so much the stuff of our everyday experience that they

have only to be stated to be recognized as obvious. Indeed, the danger is that they may be

thought to be so obvious that nothing significant can be derived from their future

examination. Yet in fact it is on the postulates of this sort that the complicated theorems of

advanced analysis ultimately depend." 31

31
Lionel Robbins, An Essay on the Nature and Significance of Economic Science
(London: MacMilan, 1935), 79 as quoted in Frederick J. Stielow, "Archival Theory Redux
and Redeemed: Definition and Context Toward a General Theory," American Archivist 54
(1991): 17.