See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228953426

Conceptual Analysis: A Method for Understanding Information as Evidence,

and Evidence as Information

Article  in  Archives and Museum Informatics · December 2004

DOI: 10.1007/s10502-005-2594-8

CITATIONS READS

116 20,337

1 author:

Jonathan Furner
University of California, Los Angeles
73 PUBLICATIONS   2,119 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Jonathan Furner on 17 March 2014.

The user has requested enhancement of the downloaded file.

Archival Science (2004) 4: 233–265
Springer 2006
DOI 10.1007/s10502-005-2594-8

Conceptual Analysis: A Method for Understanding Information

as Evidence, and Evidence as Information

JONATHAN FURNER
OCLC Online Computer Library Center, Inc., 6565 Frantz Road., Dublin, OH 43017-
3395, USA (E-mail: [email protected])

Abstract. The utility of conceptual analysis for archival science is assessed by means
of an exploratory evaluation in which the concept of evidence is analyzed. Usage of
the term ‘‘evidence’’ in the philosophies of science, law, and history is briefly
reviewed; candidates for necessary conditions of evidentiariness are identified and
examined; and taxonomies are built of evidentiariness and of archival inference.
Correspondences are shown to exist between the concepts of evidentiariness and
relevance, and between the domains of archival science and social epistemology,
thereby pointing in promising directions for further research. The tentative conclusion
is reached that conceptual analysis may profitably be used to improve understanding
of archival concepts.

Keywords: archival science, conceptual analysis, evidence, evidentiariness, inference,

information, philosophy, relevance, social epistemology

Introduction

‘‘The field of evidence is no other than the field of knowledge.’’1

The aim of this paper is to demonstrate the utility of conceptual anal-
ysis for archival science by conducting an analysis of the concept of
evidence. Conceptual analysis is a technique that treats concepts as
classes of objects, events, properties, or relationships. The technique
involves precisely defining the meaning of a given concept by identify-
ing and specifying the conditions under which any entity or phenome-
non is (or could be) classified under the concept in question. The goal
in using conceptual analysis as a method of inquiry into a given field
of interest is to improve our understanding of the ways in which

1
Jeremy Bentham, An Introductory View of the Rationale of Evidence for the Use of Non-lawyers
as well as Lawyers (ed. James Mill, 1812; first published in vol. 6 of The Works of Jeremy Bentham,
ed. John Bowring, Edinburgh and London, 1843), p. 2.
234 JONATHAN FURNER

particular concepts are (or could be) used for communicating ideas
about that field.2
Information science is one field in which important contributions have
recently been made through the application (sometimes on a relatively
informal basis) of conceptual analysis. Questions such as ‘‘What is infor-
mation?’’ and ‘‘What is a document?’’ received close attention in the
1990s, and the various suggested answers to these questions continue to
be treated as candidate cornerstones of emergent theoretical frameworks
in the field.3 In archival science, similarly, the ongoing debate about the
nature of the record received renewed impetus in the 1990s with the wide-
spread recognition that (what are called) electronic records may not share
as many of the properties of physical records as might be thought at first
glance.4 This paper seeks to demonstrate what kinds of advances might be
made possible through the application of the method of conceptual analy-
sis to key archival concepts. It does this by addressing the question ‘‘What
is evidence?’’
Two of the assumptions that underlie any positive expectation of
the utility of conceptual analysis are (i) the belief that it is at least
possible for concept-users to reach some level of agreement as to the
nature of the uses to which concepts are put, and (ii) the belief that to
reach some agreement of that kind is a prerequisite for the develop-
ment of useful (and/or interesting) knowledge (and/or theory). These
are among the assumptions that are made in the present paper. In
addition, it is assumed that any ‘‘proof’’ of the utility of the method
lies ‘‘in the pudding’’ – in other words, that it is possible to demon-
strate the utility of the method through the production of a result
that is perceived to be provocative of original or interesting ideas for
directions in which future research may be pushed. In that spirit, the

2
The body of method known simply as analysis, and deriving from the work of Frege, Moore,
and Russell, lay at the core of much twentieth-century anglophone philosophy, and its adherents
continue to exert a huge influence: see, e.g., Brian Leiter, ‘‘Introduction’’, in Leiter (ed.), The Future
for Philosophy (Oxford, 2004); and Frank Jackson, From Metaphysics to Ethics: A Defence of
Conceptual Analysis (Oxford, 1998). Analytic philosophy has a broad scope and there is no single
technique that could be claimed as the analytic method. One distinctive characteristic that is
nonetheless shared by most philosophers in the analytic tradition is the belief that many philo-
sophical problems can be illuminated by clarifying the meanings of the concepts that we use to
think about and to express those problems. This is the context in which ‘‘conceptual analysis’’ is to
be understood in this paper.
3
See, e.g., Michael Buckland, ‘‘Information as Thing’’, Journal of the American Society for
Information Science 42(5) (1991): 351–360; Buckland, ‘‘What is a ‘Document’?’’, Journal of the
American Society for Information Science 48(9) (1997): 804–809; and Michèle Tourney ‘‘Caging
Virtual Antelopes: Suzanne Briet’s Definition of Documents in the Context of the Digital Age’’,
Archival Science 3 (2003): 291–311.
4
See, e.g., Luciana Duranti et al., Preservation of the Integrity of Electronic Records (Dordrecht
and Boston, 2002).
 CONCEPTUAL ANALYSIS 235
paper concludes with two specific suggestions of lines of work that
may potentially prove productive: a collaboration among archival sci-
entists and information scientists with the aim of clarifying the nature
of the relationship between evidentiariness and relevance; and an
exploration of social epistemology as a theoretical foundation for
archival science.
Conceptual analysis, of course, is not the only method available for
clarifying or establishing evolution in archival concepts. Another plau-
sible approach to the analysis of the nature of evidence would involve
conducting a close examination of the arguments and conclusions of
others who have written about the properties, functions, and role of
evidence in archival, historical, legal, scientific, and other contexts. In
the historical context, for instance, an empirical survey could be
undertaken of the ways in which ‘‘evidence’’ has been used in the past
both by practicing historians, and by philosophers of history. Simi-
larly, in the archival context, we might wish to investigate how archi-
val practitioners and/or archival theorists have employed the concept,
and how such usage has reflected or influenced wider archival practice.
There is little doubt that studies of these kinds would be helpful in
enabling present-day theorists to reach a better understanding of what
it means for us to call certain things ‘‘evidence.’’ A preliminary effort
is made below to identify some of the more important ideas about evi-
dence that are characteristic of different fields of inquiry. The present
paper, however, is not intended as a review of prior analyses of the
concept (such as they exist), and thus no attempt is made to provide a
comprehensive set of citations to previous work.

Evidence in Theory and in Practice

The concept of evidence has historically been an important compo-

nent of ideas about the nature of the process by which people con-
struct and use arguments. Since argumentation or reasoning on the
basis of evidence is central to what has become known in the modern
period as the scientific method, it should not be surprising that philos-
ophers of science and logic have been responsible for much of the
attention that has been paid to conceptions of evidence.5 But consid-
eration of ‘‘the evidence’’ is traditionally especially important for the
practices of history and law, and thus analyses of the concept are also

5
See, e.g., Peter Achinstein (ed.), The Concept of Evidence (Oxford, 1983); and Achinstein, The
Book of Evidence (Oxford, 2001).
236 JONATHAN FURNER

relatively common in the literatures of philosophy of history and

historiography,6 and philosophy of law and jurisprudence.7
A separate but related strand of scholarship is that represented by
the attempts of theorists and practitioners in several applied fields such
as health care and public policy to establish a clear and effective link
between the findings of scientific research and the everyday practice of
working professionals. The mere claim that practice is ‘‘evidence-
based’’ seems currently to function as a convenient badge of authority;
research centers and scholarly initiatives devoted to ‘‘evidence-based
x’’ continue to spring up in diverse contexts; and a new journal, Evi-
dence and Policy, claims to be ‘‘the first [to be] dedicated to compre-
hensive and critical treatment of the relationship between research
evidence and the concerns of policy makers and practitioners.’’8
Meanwhile, steps are being taken toward the substantiation and
recognition of an interdisciplinary ‘‘evidence science’’ or ‘‘evidence
studies’’ as a distinct field of inquiry in its own right.9 Participants in
collaboratory projects include philosophers of science, philosophers of
history, statisticians, legal scholars, forensics experts, art historians,
and literary theorists, among diverse others. Most recently, the emer-
gence of the notion of ‘‘information as evidence’’ has signaled a
renewed interest among library and information scientists in the precise
nature of the relationship between conceptions of evidence and
conceptions of information.10 The present paper is delivered also as a
6
See, e.g., Arthur C. Danto, Analytical Philosophy of History (Cambridge, 1965); Carlo Ginz-
burg, ‘‘Clues: Roots of an Evidential Paradigm’’, History Workshop 9 (Spring 1980): 5–36; Michael
Stanford, The Nature of Historical Knowledge (Oxford, 1986); and Aviezer Tucker, Our Knowledge
of the Past (Cambridge, 2004).
7
See, e.g., William Twining, Theories of Evidence: Bentham and Wigmore (London, 1985);
Twining, Rethinking Evidence: Exploratory Essays (Oxford, 1990); and John Jackson and Sean
Doran, ‘‘Evidence’’, in Dennis Patterson (ed.), A Companion to Philosophy of Law and Legal
Theory (Oxford, 1996), pp. 172–183.
8
Ken Young, ‘‘Announcing a New Journal: Evidence and Policy: A Journal of Research, Debate and
Practice’’ (London, 2004), available online at http://www.evidencenetwork.org/JournalOfResearch.html.
9
See, e.g., James Chandler et al. (eds.), Questions of Evidence: Proof, Practice, and Persuasion
across the Disciplines (Chicago and London, 1994); Heather Dubrow, ‘‘The Status of Evidence’’,
PMLA: Publications of the Modern Language Association of America 111 (January 1996): 7–20;
William Twining and Iain Hampshire-Monk (eds.), Evidence and Inference in History and Law:
Interdisciplinary Dialogues (Evanston, Ill., 2003); Twining, ‘‘Evidence as a Multi-disciplinary
Subject’’ (New York, 2003), available online at http://www.ucl.ac.uk/laws/academics/profiles/
twining/evidence_multi.pdf; David Schum, ‘‘An Integrated Science of Evidence: What Kind of
Science Is It?’’ (London, 2004), available online at http://128.40.59.163/evidence/content/
schum1.pdf; and Philip Dawid, ‘‘Evidence, Inference & Inquiry: Towards an Integrated Science of
Evidence’’ (London, 2005), available online at http://www.evidencescience.org/.
10
See, e.g., Jonathan Furner et al., ‘‘Conceptions of Information as Evidence’’, in Elaine G.
Toms (ed.), ASIST 2002: Proceedings of the 65th ASIST Annual Meeting (Medford, N.J., 2002), pp.
497–498; and Anne Gilliland, ‘‘Center for Information as Evidence’’ (Los Angeles, 2004), available
online at http://www.gseis.ucla.edu/cie/.
 CONCEPTUAL ANALYSIS 237
contribution to this latter debate; it begins, however, with a brief
review of some of the terminology used in discussing related issues in
natural science, history, law, and archival science.

The Use of Evidence by Scientists

In the philosophy of science, an account of the nature of evidence is

an important component both of theories of induction and of theories
of explanation. The problem of induction is a celebrated one that has
troubled philosophers for centuries.11 Induction is the process of rea-
soning that is undertaken whenever a conclusion is inferred from a set
of premises without being entailed by those premises. In such an
argument, the set of premises forms the body of evidence that is said
to confirm (i.e., support, warrant, ground, or provide a reason to
believe) the conclusion or hypothesis. The philosophical ‘‘problem’’ of
induction, as it is typically conceived, consists in providing a convinc-
ing, noncircular justification for making inferences of this kind – i.e.,
an explanation (of why it is rational for us to accept conclusions
reached by induction) that does not itself rely on an inductive
argument.
Before it can offer such a justification, however, any theory of
induction needs to provide a clear and full description of the type of
process that it will go on to explain. Description is more difficult
than might at first be assumed, since the forms of nondeductive
arguments are numerous and various. Norton12 groups candidate
descriptions of induction (sometimes known as models of confirma-
tion) into three families. Models in one family treat induction as a
process of generalization, whereby we infer from observed evidence
that some As are B to the hypothesis that all As are B. The evidence
e is said to confirm the hypothesis h as a corollary of its being an
instance of h. We might consider our observation of a white swan,
for example, to be evidence for the hypothesis that all swans are
white. Models in a second family treat induction as a hypothetico-
deductive (H-D) process, whereby we make a conjecture of a hypothe-
sis, make a prediction of an event that is entailed by that hypothesis,
and make an observation of the occurrence or nonoccurrence of that
event. The event or evidence e is said to confirm the hypothesis h as
a corollary of its being entailed by h. For example, we might make
11
See, e.g., Ian Hacking, An Introduction to Probability and Inductive Logic (Cambridge, 2001).
12
John D. Norton, ‘‘A Little Survey of Induction’’, in Peter Achinstein (ed.), Scientific Evidence:
Philosophical Theories and Applications (Baltimore, Md., forthcoming).
238 JONATHAN FURNER

the conjecture that long-term cigarette smoking causes lung disease. If

this hypothesis were true, then we would expect a patient who had
been smoking over a long period to have contracted lung disease.
This expectation would be the result of logical deduction from the
hypothesis. If we then found our expectation to match our observa-
tion of a particular patient, then our hypothesis would be confirmed
by that evidence. Models in a third family treat induction as a proba-
bilistic process, whereby we calculate the effect of observed evidence
on our degree of belief in a hypothesis. The evidence e is said to con-
firm the hypothesis h as a corollary of p(h|e)>p(h), where p(h) repre-
sents our degree of belief in h before our observation of e, and p(h|e)
represents our degree of belief in h after our observation of e. We
might, for example, hypothesize that Edward de Vere was the real
author of the dramatic works commonly attributed to Shakespeare.
Prior to considering evidence of the stylistic similarities between de
Vere’s surviving letters and Shakespeare’s plays, we might be skepti-
cal; after considering such evidence, our degree of belief might rise
considerably, in which case we would say that the evidence confirms
the hypothesis – or, at least, confirms it to an extent that corresponds
to the increase in the degree to which we now believe it. This focus
on degrees of belief allows for the insertion, into probabilistic
descriptions of the inductive process, of a subjective element that is
conspicuously absent from other approaches.
Models in the H-D family may be distinguished from one another
by the different responses they exemplify to the problem of underde-
termination. A hypothesis is said to be underdetermined in any case in
which the observed evidence is entailed by more than one hypothesis.
The observation of the smoker with lung disease is consistent not
only with the hypothesis that smoking causes lung disease but also
(for instance) with the hypothesis that lung disease somehow engen-
ders a craving for cigarettes. Since it would seem that, in fact, every
case is of this kind, any model of confirmation must show how we
select from among competing hypotheses. One variant on the basic
H-D model is that of abduction, or inference to the best explanation
(IBE), which treats induction as a process of coming to identify the
hypothesis that serves as the ‘‘best’’ (i.e., simplest, most coherent,
most useful, most elegant ...) explanation for the observed evidence.13
A related subclass of variants includes reliabilist accounts that
propose, as a criterion for hypothesis selection, the degree of past

13
See, e.g., Peter Lipton, Inference to the Best Explanation (London and New York, 2nd edn.,
2004).
 CONCEPTUAL ANALYSIS 239
reliability of the methods by which individual hypotheses are reached.
If, after examining past performance, we were to find that, given a
choice between the explanation that posits the existence of the fewest
new or unobserved classes of object and the explanation that covers
the largest number of cases, selection of the explanation of the former
kind appears to reflect reality more frequently, then on the reliabilist
account we would be justified in using that method of deciding
among competing hypotheses. Of course, that smoking causes lung
disease is widely considered among health professionals as the ‘‘best’’
explanation of the unhealthy condition of many smokers: it may be
no coincidence that it is also the simplest.
If it is to be persuasive, any model of IBE needs also to provide
an account of the process of explanation itself; typically, in this
context, a causal account of explanation is given that equates expla-
nation with specification of a historical sequence of cause and effect.
Such an account serves as an alternative to the well-known
deductive-nomological (D-N) model, whereby to explain any given
event is to deduce it from a ‘‘covering law’’ or generalization.14
According to the D-N model, explaining the incidence of lung dis-
ease in a smoker is a matter of stating the ‘‘law’’ that long-term
smoking is always accompanied or followed by lung disease, and
then deducing the facts of the particular case from that law. The
D-N model has been criticized for its inadequacy on several fronts,
not least for its underdetermination of the class of acceptable expla-
nations for any given event. The particular ‘‘law’’ stated above,
much like the proposition that all swans are white, is not empiri-
cally true, of course; yet it is another feature of the D-N model that
any putative law can be modified by adding exception clauses (such
as ‘‘... except in cases where patients have condition x’’) as neces-
sary to cover all cases that arise. It is notable that historiographers
and philosophers of social science who have considered the D-N
model as it has been developed by their colleagues in philosophy of
natural science have widely rejected it in favour of accounts of
explanation that do not require deterministic covering laws to be
specified for human behavior, but that allow for the effects of
human intentionality, individuality, and the exercise of free will, and
that require the explainer to engage in imaginative interpretation of
the meanings events have for human actors.15

14
See, e.g., Carl G. Hempel, Aspects of Scientific Explanation and Other Essays in the Philosophy
of Science (New York, 1965).
15
See, e.g., William H. Dray, Philosophy of History (Englewood Cliffs, N. J., 1964).
240 JONATHAN FURNER

The Use of Evidence by Lawyers

In a legal context, evidence is information that is presented to a judge

or jury at a trial in order to prove or disprove a given fact – that is,
in order to convince the judge/jury of the truth or falsity of that
fact.16
The oral reports of events made by eye-witnesses while under oath
to tell the truth are normally construed as constituting direct evi-
dence, in the sense that (given certain assumptions about the authen-
ticity, reliability and accuracy of the witness) judge or jury need make
no further inference from the content of such testimony in order to
learn the facts of the case. Testimony may be presented in documen-
tary rather than oral form; the category of documentary evidence can
include images (photographs, drawings, videos, etc.) and recordings of
speech as well as written texts.
Circumstantial evidence is evidence of circumstances from which a
judge/jury may reasonably infer a fact that cannot be proven directly.
For instance, a fingerprint left at the scene of a crime could be trea-
ted as circumstantial evidence for the past (unwitnessed) presence at
the scene of a person known to be capable of leaving the print in
question. Circumstantial evidence is often presented in physical form;
the category of real evidence includes artifacts and natural objects
(e.g., items on which fingerprints or footprints are left, or that are
damaged or tampered with) whose significance lies in their bearing
traces of human activity of a nonverbal and nongraphical (i.e., non-
documentary) nature.
So that the right of an individual to a fair trial is protected,
only evidence of certain kinds is allowed or admissible in a trial.
Hearsay (i.e., secondhand testimony), and evidence obtained by ille-
gal search, for example, are typically deemed inadmissible. Rules of
evidence specify how admissible evidence is to be distinguished from
inadmissible. In the United States, federal courts follow the Federal
Rules of Evidence; state courts have developed their own sets of
rules.
Modern understanding of the different types of evidence used in
civil and criminal justice, and the different methods to be used in
obtaining and interpreting evidence, relies to a great extent on the
continuing influence of two seminal works: Bentham’s17 Rationale of

16
See, e.g., Christopher B. Mueller and Laird C. Kirkpatrick, Evidence under the Rules: Text,
Cases, and Problems (New York, 1996).
17
Jeremy Bentham (1748–1832), British philosopher, economist, and jurist.
 CONCEPTUAL ANALYSIS 241
Judicial Evidence (1827)18 and Wigmore’s19 Science of Judicial Proof
(1937).20 Twining21 and Shapiro22 trace the history of the develop-
ment of legal notions of evidence; Binder & Bergman23 and Anderson
and Twining24 provide accounts of the process of establishing legal
proof; Tillers and Green25 and Schum26 use the framework provided
by Bayesian probability theory to model this process.

The Use of Evidence by Historians

The concept of evidence as used in archival science is one that is

closely related to that used by historians. In this respect, it should be
noted that conceptual analysis of the kind that was characteristic of
twentieth-century analytical philosophy is hardly in vogue in contem-
porary historiography. Questions like ‘‘What is historical knowledge?’’
and ‘‘What is historical explanation?’’27 now seem to belong to a dif-
ferent age – not least, of course, because much work in the philosophy
of history over the last 40 years has led many to doubt that historical
inquiry can produce anything that may appropriately be conceptual-
ized as objective knowledge or explanation.28 Archival theorists of a
postmodernist bent might understandably be suspicious of any mode
of inquiry that appears to seek fundamental essences or absolute
truths. Given the direction of recent trends in archival theory,29 it
18
Jeremy Bentham, Rationale of Judicial Evidence, Specially Applied to English Practice, 5 vols.
(ed. John Stuart Mill, London, 1827).
19
John Henry Wigmore (1863–1943), American legal scholar.
20
John Henry Wigmore, The Science of Judicial Proof (3rd edn. of The Principles of Judicial
Proof as given by Logic, Psychology and General Experience and Illustrated in Judicial Trials,
Boston, 1937).
21
Twining, Theories of Evidence, already cited.
22
Barbara J. Shapiro, ‘‘Beyond Reasonable Doubt’’ and ‘‘Probable Cause’’: Historical Perspec-
tives on the Anglo-American Law of Evidence (Berkeley, California, 1991).
23
David A. Binder and Paul B. Bergman, Fact Investigation: From Hypothesis to Proof (St. Paul,
Minn., 1984).
24
Terence Anderson and William Twining, Analysis of Evidence: How to Do Things with Facts
Based on Wigmore’s Science of Judicial Proof (Boston and Toronto, 1991).
25
Peter Tillers and Eric D. Green (eds.), Probability and Inference in the Law of Evidence: The
Uses and Limits of Bayesianism (Boston and Dordrecht, 1988).
26
David A. Schum, The Evidential Foundations of Probabilistic Reasoning (New York, 1994).
27
Questions of this kind are central for, e.g., W. H. Walsh, An Introduction to Philosophy of
History (3rd edn., London, 1967); and Michael Stanford, An Introduction to the Philosophy of
History (Malden, Mass., and Oxford, 1998).
28
See, e.g., C. Behan McCullagh, The Truth of History (London and New York, 1998); and
Alun Munslow, The New History (Harlow, England, 2003).
29
See, e.g., Terry Cook, ‘‘Archival Science and Postmodernism: New Formulations for Old
Concepts’’, Archival Science 1(1) (2001): 3–24; and Eric Ketelaar, ‘‘Tacit Narratives: The Meanings
of Archives’’, Archival Science 1(2) (2001): 131–141.
242 JONATHAN FURNER

might be appropriate to provide a brief, preemptive defense of my

decision to focus on a technique that seems to so carefully avoid con-
fronting some of those issues – issues to do with, for example, the
very possibility of objectivity, of a distinction between fact and value,
or of one between language, thought, and reality – that (as recent
work in historical studies has taught us) are among the most impor-
tant for scholars in all fields. The following is presented in that vein.
To summarize: It is sometimes thought possible to distinguish, at
the grossest of levels, between those historians whose philosophy,
once rendered explicit, might be characterized as ‘‘positivist,’’ and
those whose approach is classifiable as ‘‘postmodernist.’’ The gener-
ality of these simplistic labels, and the tendency for each of them to
be used pejoratively, militate against their utility. What can neverthe-
less be productive is to identify some of the more important episte-
mological assumptions that, when considered collectively, could
plausibly be used as the basis on which these generic labels are
assigned.
A stereotypically positivist approach to historical inquiry rests on
assumptions of the following kinds:

There is an external reality that exists independently of human
thought.

The past is a component of reality consisting of a series of events
that have already occurred at particular times and in particular
places.

There is a single objective description or representation of reality
(and, thus, the past) consisting of a set of statements – the facts –
that are true.

The truth of an individual statement is the degree to which it cor-
responds to reality.

It is possible to have knowledge about reality – i.e., to make state-
ments about reality that can independently be shown to be true
(or, at least, to resist falsification).

The only method by which reality may be known is the scientific
method of inductively testing hypothetical generalizations (deter-
ministic or probabilistic) against empirical observation of individ-
ual events – the evidence.

Scientists (including historians) are neutral observers and gatherers
of true, objective descriptions of events.

The occurrence of future events can be explained (i.e., predicted)
– and past events similarly retrodicted – by identifying initial
conditions and hypothetical generalizations that serve as
 CONCEPTUAL ANALYSIS 243
premises from which the occurrence of particular events can be
deduced.

The practice of history is analogous in this respect to that of natu-
ral science.
In contrast, a stereotypically postmodernist approach might run as
follows:

There is no external reality that exists independently of human
thought. What is called ‘‘reality’’ is a construction of human
thought.

There is no single description or representation of reality that is
true. What is called ‘‘the facts’’ at any given point is that set of
statements endorsed by the group of people most well equipped to
impose their values over others.

There is no such property as truth. An evaluation of a statement
as true by any given person indicates only that person’s positive
attitude toward it.

It is not possible to have knowledge about reality (past, present,
or future).
Of course, between the extremes represented by these two incom-
mensurable caricatures lies a wide and diverse range of alternative
perspectives. One group of relatively common ‘‘third ways’’ consists
of those that accept the reality of an external world, but that insist on
the impossibility of our ever knowing that reality ‘‘as it is in itself,’’
since we are all necessarily trapped within our own perspective. Far
from acting as an encumbrance, however, it is this apparent limitation
that makes the practice of history possible; for, if historians were
really the neutral observers of an objective truth that the positivist
paradigm requires, they would be unable to evaluate the relative sig-
nificance of past events, or to make selections from sources of
potential evidence, or to interpret the thoughts, feelings, and inten-
tions of human agents. On this basis:

The truth of an individual statement is the degree to which it is
perceived to cohere with other true statements.

It is possible to obtain an understanding of the past through the
hermeneutic method of interpreting people’s thoughts, feelings,
and intentions.

The practice of history is analogous in this respect to writing
fiction, and its products – narratives – are products of the histo-
rian’s imagination.
244 JONATHAN FURNER

As Fay30 points out, the pendulum of scholarly thinking has

oscillated from the ‘‘Scientific’’ (or positivist) to the ‘‘Rhetorical’’
(or postmodernist) attitude many times in the last five hundred
years, and there is no particular reason to think that the contempo-
rary manifestation of the ‘‘linguistic turn’’ in philosophy of history
(as represented by the content of, for example, History and Theory
since that journal’s inception in 1960) will be the last such swing.
Moreover, and more immediately, it is important to recognize that,
notwithstanding the insights supplied by the ‘‘Rhetorical’’ attitude –
viz., most significantly, recognition of the necessarily subjective
nature of any historian’s perception of reality, selection of evidence,
and representation of the facts – many practicing historians continue
to work as if their goal is to get closer to a (if not the) truth, and
to persuade their peers not only of the validity of their arguments
but also of the truth of their conclusions. Successful persuasion is
often the result of the application of skills such as the ability to
empathize with historical actors or to construct a powerful narra-
tive; but equally as often (and typically simultaneously) such success
is reliant on the historian’s demonstration that conclusions are
justified by means of appeals to (what is taken to be) the evidence,
and revised in the light of new evidence. So long as historians’
products continue to be evaluated in this manner, it would seem to
be appropriate to continue to seek clarification as to the ways in
which evidence is (or could be) conceptualized in contexts of this
kind.

The Use of Evidence by Archivists

Both historians and lawyers are interested in evidence primarily for

the light that it may throw on events that occurred (or situations that
existed) in the past. As those responsible for the selection, organiza-
tion, and preservation of artifacts of various kinds whose existence or
content may be construed as (or as yielding) evidence of past events,
archivists and archival scientists have also been concerned to clarify
understanding of the concept.31

30
Brian Fay, ‘‘The Linguistic Turn and Beyond in Contemporary Theory of History’’, in Fay
et al. (eds.), History and Theory: Contemporary Readings (Malden, Mass., and Oxford, 1998),
pp. 1–12.
31
See, e.g., Heather MacNeil, Trusting Records: Legal, Historical, and Diplomatic Perspectives
(Dordrecht and Boston, 2000); and Brien Brothman, ‘‘Afterglow: Conceptions of Record and
Evidence in Archival Discourse’’, Archival Science 2(3/4) (2002): 311–342.
 CONCEPTUAL ANALYSIS 245
Twentieth-century thought on the nature of archival evidence
demonstrates the distinctive influence of the two doyens of modern
archival practice, the Briton Hilary Jenkinson (1882–1961), and the
American Theodore R. Schellenberg (1903–1970).32 Jenkinson popu-
larized the idea that evidentiariness is the essence of archival
records, and insisted on the universal, objective strength of the link
between premise (the existence of the record) and conclusion (the
occurrence of the events that produced the record); Schellenberg
articulated a much-repeated distinction between the evidential and
the informational value of a record that continues to inform contem-
porary archival theory and education. For Schellenberg, a record’s
evidential value is an index of its utility in documenting the circum-
stances of its creation, whereas its informational value reflects the
importance of its symbolic content. The possibility of confusion is
thus introduced by Schellenberg’s use of the term ‘‘evidential’’ to
refer only to evidence of a real (rather than documentary) kind. In
historical scholarship and in the law, Schellenberg’s distinction is an
important one, but it is one that is drawn between different kinds
of evidence (i.e., between real evidence-as-trace and documentary
evidence-as-testimony) rather than between evidential and non-
evidential values.
It is well appreciated that Jenkinson’s views on the essential objec-
tivity and truthfulness of the archives as a record of ‘‘what really
happened’’ in the past reflects the positivist, Rankean historiography
that was dominant in his time.33 For Jenkinson, appraisal of records’
long-term value was a task that should be undertaken only by
records creators, and that should involve consideration only of the
functions of records as originally intended. Schellenberg’s supporters
argued, in contrast, that appraisal decisions should take into account
any expectations as to the future use of records for purposes other
than those for which they were created, and that archivists rather
than records creators should bear responsibility for decisions of that
kind.
In the late twentieth century, and primarily in Australia and Can-
ada, a neo-Jenkinsonian perspective on archival best-practice emerged
that retained Schellenberg’s emphasis on appraisal, but that restored
to primacy Jenkinson’s principle that selection decisions are to be
based on evaluations of the strength of the relationships between
32
See, e.g., Terry Cook, ‘‘What Is Past Is Prologue: A History of Archival Ideas Since 1898, and
the Future Paradigm Shift’’, Archivaria 43 (Spring 1997).
33
Leopold von Ranke (1795–1886), German historian renowned for his dictum that history
should be recorded ‘‘wie es eigentlich gewesen.’’
246 JONATHAN FURNER

records and the contexts in which they were originally created. The
scope of a record’s contextuality is now more broadly defined than
Jenkinson allowed. In Canada, Hugh Taylor and Terry Cook are
among those who have promoted a new focus on the ‘‘macro,’’ socie-
tal contexts that give records their meaning.34 One result of this trend
is for the primary role of the archivist to now be more commonly
perceived as one of determining the provenance of records as a means
of understanding the social structures and processes by whose com-
plex combination records are generated. In Australia, Sue McKem-
mish and Frank Upward have been concerned to emphasize the
importance of archives as the primary means by which societies main-
tain the accountability of their institutions.35 In this sense, the process
by which records of demonstrable authenticity provide evidence of
public and private institutions’ actions is an essential component of
any functioning democracy.
More recently, McKemmish and her colleagues have begun more
explicitly to explore the similarities and differences between corporate
recordkeeping, where records are captured as evidence of business
functions and activities, and personal recordkeeping, where records are
captured as evidence of the roles and activities of individuals.36 Just as
archivists have long appreciated the importance of establishing record-
keeping and preservation systems that can capture, manage and
maintain complete and accurate records of business activities for busi-
ness purposes, McKemmish is concerned to emphasize the active role
of archivists in institutionalizing personal recordkeeping systems as a
means of preserving the collective memory (and thus the identity) of
societies and cultures. No redefinition of ‘‘record’’ is required or
implied in this work: the recordness of a document such as a personal
letter is still conceived as consisting in its transactional origin (as the
product of some human interactivity), and its evidential potential (as
the object of a recordkeeping process that maintains the document’s
links to the context in which it was generated); and the functional
requirements of any recordkeeping system are still conceived as
including the means to guarantee the authenticity and reliability of
documents captured as evidence.
34
See, e.g., Tom Nesmith, ‘‘Hugh Taylor’s Contextual Idea for Archives and the Foundation of
Graduate Education in Archival Studies’’, in Barbara L. Craig (ed.), The Archival Imagination:
Essays in Honour of Hugh A. Taylor (Ottawa, 1992), pp. 13–37; and Terry Cook, ‘‘Mind Over
Matter: Towards a New Theory of Archival Appraisal’’, in Craig (ed.), The Archival Imagination,
pp. 38–70.
35
See, e.g., Sue McKemmish and Frank Upward (eds.), Archival Documents: Providing
Accountability Through Recordkeeping (Melbourne, 1993).
36
See, e.g., Sue McKemmish, ‘‘Evidence of Me ...’’, Archives and Manuscripts 24(1) (1996): 28–45.
 CONCEPTUAL ANALYSIS 247
General Characteristics of Evidence

From the foregoing, it may be determined that ‘‘evidence’’ may be

generally and loosely defined as that which we consider or interpret in
order to draw or infer a conclusion about some aspect of the world.37
Correspondingly, something may be identified as being evidential or
evidentiary if it has been (or may potentially be) considered or inter-
preted with the purpose of drawing a conclusion. As a hypothetical
case – one to which we shall return several times in the course of this
paper – we might talk of the ‘‘evidence’’ supplied by a physical artifact,
perhaps one recently found in an archaeological dig, for any statement
that we may wish to make about the cultural context in which that
artifact was produced. Such an artifact would be deemed to be ‘‘evi-
dentiary’’ simply in virtue of its playing the particular kind of role that
we have decided it is to play in our argument. Suppose that the arti-
fact in question has the appearance of a stone vase and that its outer
surface is inscribed with a sequence of symbols. If, following an exami-
nation of the artifact, we were to draw conclusions – conclusions about
the artifact, about its functions and/or meanings, about the contexts
in which it was created, about the classes of objects of which it is a
member – it might seem perfectly reasonable for us to refer to the
artifact itself as ‘‘the evidence,’’ that is, the object of interpretation
that prompted us to make our inferences. However, this usage of the
term reflects just one of various senses in which the concept of
evidence may usefully be employed in this context.
As a point of entry to a more detailed examination of the structure
of cases such as that outlined above, I would like to make a number of
preliminary observations about the concept of evidence by considering
the various ways in which the term is used in ordinary discourse.

Evidence is relational

We commonly speak of things as being evidence of or for other

things. On this basis, we may wish to define evidentiality or
37
The English noun ‘‘evidence,’’ like its equivalents in French (e´vidence), Italian (evidenza), and
Spanish (evidencia), derives from the Latin combination of the prefix e-meaning ‘‘out’’ and the verb
videre meaning ‘‘to see.’’ The compound verb evidere, literally ‘‘to see out,’’ means ‘‘to look’’ in the
sense of ‘‘to appear to be [well, tired, happy, etc.].’’ To be evident is to be visible, obvious, or
certain. According to the Oxford English Dictionary, the two primary meanings of ‘‘evidence’’ in
English are (1) the quality, condition, or state of being evident, and (2) that which manifests or
makes evident. The latter is the sense with which we are concerned in this paper. Two sub-senses of
this second sense are (a) a sign from which inferences may be drawn, and (b) a ground for belief.
The distinction between these two sub-senses is dealt with at a later point in the paper.
248 JONATHAN FURNER

evidentiariness as that type of relationship that is conceived as linking

x and y, when x is treated as evidence for y. Sometimes, and poten-
tially confusingly, the word ‘‘evidence’’ is itself used instead of ‘‘evi-
dentiality’’ to refer to the relationship (or type of relationship)
between x and y, rather than (or as well as) to refer to x itself.
What is it possible for x and y to be? Are they instances of the
same kind of thing, or of different kinds? In the general definition of
evidence given above, we indicated that y could be conceived as a
conclusion, somehow to be inferred. In order to clarify this general
conception of evidence-based reasoning, it is first necessary to explain
what is meant here by the terms ‘‘infer’’ and ‘‘conclusion.’’ And to do
even that, it is necessary to bring to the surface two important
assumptions that are often made only implicitly in discussions of this
kind – one about the essential duality of the physical and mental
worlds (or of reality and representation), and another about the
difference between particular mental entities and classes of mental
entities.
A common assumption that guides much of our practical activity
in the world is that it is possible to distinguish between, on the one
hand, the real situations that we encounter and the real events that we
experience in the physical world, and, on the other, any ideas, beliefs,
or thoughts we might have about those situations and events.38 Fur-
thermore, it would seem useful to identify a distinction between the
particular thought-tokens that make up each of our own personal
mental states, and the thought-types that are instantiated by those
tokens. Your thought that the battle of Hastings was fought in 1066
and my thought that it was fought in that year are two separate
instances of the same thought-type. A convenient way of expressing
this latter distinction is to refer to thought-tokens simply as thoughts,
and to thought-types as propositions.
An inference – an argument or process of reasoning – is made up
of a series of propositions. The final proposition to which the
argument leads is known as the conclusion; the propositions that are
analyzed in the lead-up to the conclusion, and from which the conclu-
sion derives, are known as the premises.39 A conclusion, in other
words, is an idea or proposition about the world that is the product
of a process of inferential analysis.

38
The existence of a distinction between ‘‘reality’’ and any representation of reality is denied by
the rigorous postmodernist, who would treat all entities as thoughts (in the sense used here).
39
In a legal context, the conclusion might be referred to as ‘‘the point in question;’’ in history, as
‘‘the facts;’’ and in science, as a hypothesis.
 CONCEPTUAL ANALYSIS 249
How are conclusions derived from premises? What methods are
used in the inferential analysis of premises? We may distinguish
broadly between two kinds of inferential method: deduction and
induction. A simple example of a deductive inference is provided by
the following argument, represented in three systems of notation. In
this example and the next one, the proposition listed below the line is
the conclusion; the propositions listed above the line are the premises.

Pl If Xi is a vase made of stone, If Xi is a, then Xi is b. p fi q.

then Xi was made in Ireland. (Or,
Everything that is a vase made of
stone was made in Ireland; or, All
vases that are made of stone were
made in Ireland.)
P2 Vase 1 (the artifact in question) is a X1 is a. p.
vase made of stone.

C1 [ Vase 1 was made in Ireland. [ X1 is b. [ q.

If we can demonstrate the truth of the premises here, we thereby

demonstrate the truth of the conclusion. Note, however, that an eval-
uation of the validity of the deduction does not depend on an evalua-
tion of the truth of either of the premises, but is rather an evaluation
of the form of the argument. The general form instantiated here is
known in the propositional calculus as modus ponens. The expression
p fi q is called a conditional, in which p is the antecedent and q the
consequent. If p fi q is true, then p is a sufficient condition for q: we
would need only to know that the artifact in question is a stone vase
for us to safely conclude that it was made in Ireland. Moreover, if
p fi q is true, then q is a necessary condition for p: we would need
only to know that the artifact in question was not made in Ireland
for us to safely conclude that it is not a stone vase.

P1 Vase 2 is made of stone and X2 is a and X2 is b. p  q.

was made in Ireland.
P2 Vase 3 is made of stone and X3 is a and X3 is b. p  q.
was made in Ireland.
Pn ...Vase n is made of stone and ...Xn is a and Xn is b. ...p  q.
was made in Ireland.

C1 [ If Xi is a vase made of stone, [ If Xi is a, then Xi is b. [ p ! q.

then Xi was made in Ireland.
250 JONATHAN FURNER

An example of an inductive inference, represented in the same three

systems of notation, may be given as above. Clearly, our observation
of only one counter-example (of a vase that is made of stone but not
made in Ireland) would be sufficient for us to be 100% certain of
the falsity of the conclusion as stated here. But we can never be 100%
certain of its truth, since the inference is not a matter of simple
deduction, but rests on the additional assumption that the non-
occurrence of counter-examples in the past is a reliable indicator of
their non-occurrence in the future.

Evidence is probabilistic

The weight of evidence may be measured

In practice, many arguments take a probabilistic form of a kind that
merges inductive and deductive inference as in the following example:

P1 Vase 2 is made of stone and was made in Ireland.

P2 Vase 3 is made of stone and was made in Ireland.
Pn ...Vase n is made of stone and was made in Ireland.

C1/P3 [ If Vase i is a vase made of stone, then it is likely that

Vase i was made in Ireland.
P4 Vase 1 is made of stone.

C2 It is likely that Vase 1 was made in Ireland.

It is common in this context to speak of the weight of evidence

supporting a given conclusion. The assumptions made are (i) that the
relative frequency of occurrence of past events in similar circum-
stances can reasonably serve as a predictor of the likelihood with
which such events will occur in the future, and (ii) that the amount,
extent, or degree of support or warrant that a piece of evidence sup-
plies for the assertion of or belief in a given conclusion may be
weighed, measured, or quantified by calculating relative frequencies or
probabilities of that kind.

The relevance of evidence may be measured

The use of the phrase ‘‘in similar circumstances’’ in the previous para-
graph is highly significant. The exact position of the threshold level of
similarity that pairs of contexts must exceed for them to be consid-
ered relevantly related in this sense is a matter of subjective opinion
that nevertheless has great bearing on the relevance of certain pieces
of evidence. Suppose, for example, we come across another vase that
 CONCEPTUAL ANALYSIS 251
is made of stone but that we determine was not made in Ireland.
Would we thenceforth treat that vase as evidence against drawing
conclusion C2? Not necessarily – not if it becomes clear that there is
something different about that vase or about the context in which it
was produced that we decide should prevent it from being considered
as a member of the class of relevantly related vases or contexts. We
might recognize, perhaps, that the new vase, whilst made of stone, is
made of a slightly different kind of stone from all the others – in
which case, rather than taking its existence into account as evidence
for rejecting our original conclusion about Vase 1, we might be led
simply to revise our premises in a way that denies its relevance.

Evidence can be mental (or ‘‘ideal’’); evidence can be physical

(or ‘‘real’’)

All the components of the inferences illustrated so far may be catego-

rized as mental entities. They are ideas about situations in the physical
world; they are not those situations themselves. It is important to
bear this in mind when evaluating any taxonomy of kinds of evi-
dence. Just as we often speak of mental entities – things that exist in
our minds, such as ideas, thoughts, and beliefs, and classes of such
entities, such as propositions – as being evidence, we also often speak
of events, objects, or situations in the physical world external to our
individual minds as being evidence (or sources of evidence). Some-
times the word ‘‘evidence’’ is used to cover entities of both kinds;
sometimes it is used more narrowly to include entities of only one
specific kind. For example: Rather than talking of the proposition p,
that our vase is made of stone, as evidence, we might wish instead to
talk of the vase itself (or, perhaps, its having the property of being
made of stone) as evidence.

Evidence can be substantive; evidence can be attributable

So, it should also be clear that not only do we commonly speak of

things as being evidence, we also speak of things – sometimes the
same things or kinds of things – as being sources of evidence, as con-
taining, providing, or supplying evidence, or simply as having evi-
dence. In talking about our hypothetical case, we might plausibly
decide that it is somehow more precise to treat the vase as a source of
evidence, rather than to treat it as evidence itself. The vase as source,
and the evidence supplied, are thereby kept conceptually distinct. In
such a case, it is typically possible to distinguish between the mental
252 JONATHAN FURNER

evidence on the one hand, and the physical sources of that evidence
on the other. We might, for instance, talk of our vase as being the
physical source of an idea that we have about the context in which it
was produced. In recognition of this distinction, we should be pre-
pared also to acknowledge that the word ‘‘evidence’’ may sometimes
be used loosely to denote a source or sources, rather than to refer to
that which we might more properly conceive as the evidence itself.
What is more common is the description of sources as ‘‘evidentiary,’’
i.e., as supplying evidence.

Evidence can be form; evidence can be content

Some events and objects in the real world may be identified as human
artifacts in that they are generated as a result of human activity.
Some of these artifacts may be treated as documents in virtue of their
having (symbolic) content as well as (physical) form. Sometimes such
documents are treated as evidence (or sources of evidence) in virtue of
their form; sometimes documents are treated as evidence in virtue of
their content. In this sense, content is conceived as anything that
serves as the physical expression or representation of human ideas or
thoughts. The relationship of content to thought is that of meaning.
In any given case, this relationship is complex, not least because any
instance of content (i.e., an utterance) may have multiple potentially-
associated thoughts, including at least (i) a meaning intended by the
speaker or writer, (ii) a conventional meaning emergent from prior
usage, and (iii) a meaning assigned by the hearer or reader, none of
which may coincide.
It should be clear, for example, that vases of the kind observed in
our hypothetical case can be treated as sources of evidence in virtue
of their form. We have already seen how an argument may be con-
structed in which the known origins of previously-examined vases are
used as evidence in support of a conclusion about the origins of a
newly-discovered vase. This kind of argument, in which propositions
about the form of other vases are marshaled as evidence in support of
an answer to a question about the context in which the vase in ques-
tion was produced, is (as we shall see) just one of the kinds of infer-
ence in which evidence-of-form may play a part. Another is that in
which the very existence of an artifact is invoked as evidence of the
occurrence of the act that produced it.
But our vase and others like it can also be treated as sources of
evidence in virtue of their content. The series of symbols inscribed
on Vase 1’s outer surface comprise an utterance, or speech act, carried
 CONCEPTUAL ANALYSIS 253
out by an author or inscriber. Such an utterance expresses a
thought or particular idea of the author’s (its intentional meaning),
and may also be assigned meaning by individual readers or through
inter-subjective consensus or convention. Documents – i.e., artifacts
that, like our vase, have symbolic content – supply evidence not just
of the occurrence of the acts that produced them, but of the occur-
rence of the acts that they document. Suppose that, on interpreting
the inscription on our vase, we were to discover a story about an
ancient battle. Not only would we then have evidence of the occur-
rence of the speech act of which this document is a result, but we
would have evidence of the occurrence of the battle described. The
need to evaluate both the authenticity of the document and the reli-
ability of the author if we are to quantify the weight of the evidence
in support of these conclusions is discussed in a separate section
below.

Evidence can be attitudinal; evidence can be propositional

In taxonomies of mental entities, a distinction is commonly made

between propositions, which have truth-value or epistemic status, and
beliefs, which may be treated as personal evaluations of the corre-
spondence to reality of given propositions, or as preferential atti-
tudes to those propositions. Thus, the general proposition that the
vase in our example is red in color is distinguishable from my per-
sonal belief that that proposition is true. Similarly, on this reading,
the general proposition that our vase was produced in Ireland may
be distinguished from the particular belief that any individual may
have that the vase was, in reality, produced in Ireland. Beliefs
(along with proposition-targeted thoughts or attitudes of other kinds
such as hopes, intentions, etc.) may be viewed as individual instan-
ces of propositions that may themselves be viewed as thought-types
or thought-classes.

A Taxonomy of Kinds of Evidentiariness

Where does the preceding discussion of the general characteristics of

evidence get us? Is it possible to establish a taxonomy of kinds of
evidentiariness that further clarifies our understanding of what it is to
be evidentiary? The following is presented as an effort to impose some
conceptual order on the matter. In this taxonomy, two general
254 JONATHAN FURNER

categories of evidentiariness are identified: evidentiariness as a

property of sources; and evidentiariness as a property of ideas.40

Evidentiariness as a property of sources

Evidentiariness as a property of situations and events,

or of their attributes
We might say that a situation, made up of a set of physical objects
and their relationships, or an event, a change in an existing situation,
is what is evidentiary. In our hypothetical case described above, it is
the vase itself that on this reading would be identified as having evi-
dentiariness. Alternatively, but relatedly, we might argue that it is the
various individual properties, attributes, or characteristics of the vase
– for example, its existing, or its being made of stone – that should be
treated as having evidentiariness.

Evidentiariness as a property of utterances

Our inscribed stone vase is not simply an artifact. It is a documentary
artifact, or document, in virtue of its having a symbolic component.
The series of symbols inscribed on its surface comprise an utterance,
or speech act, carried out by the author or inscriber. We may reason-
ably decide to treat the utterance itself – the sequence of physical sym-
bols to which meaning may be assigned – as that-which-is-evidentiary.

Evidentiariness as a property of ideas

Evidentiariness as a property of ideas about form

Instead of identifying evidentiariness as a property either of physical
things or of attributes of physical things, we might wish to say that
propositions or ideas about particular situations, events, or their for-
mal attributes – for example, the idea that the vase exists, or the idea
that the vase is made of stone – are what are evidentiary. Such ideas
may be derived from our direct observation or experience of the situ-
ation or event in question.

Evidentiariness as a property of ideas about content

Alternatively, we might wish to consider as evidentiary the proposi-
tions or ideas about particular situations, events, or their attributes
that may be assigned to the utterances or symbolic components of
40
In the Oxford English Dictionary, among the listed senses of both ‘‘evidentiary’’ and ‘‘evi-
dential’’ are these: ‘‘furnishing [i.e., supplying] evidence,’’ and ‘‘having the nature of [i.e., being]
evidence.’’ The former corresponds to the sense in which evidentiariness is a property of sources,
the latter to that in which evidentiariness is a property of ideas.
 CONCEPTUAL ANALYSIS 255
documents. Thus, we would treat the meaning(s) that we assign to
the inscription on the outer surface of the vase as that-which-is-
evidentiary.
It is possible to invoke any of these senses in amplification of the
broad definition that was introduced earlier, of evidence as that which
we consider or interpret in order to infer a conclusion. Although care
was taken above to define both deduction and induction as inferential
processes that begin and end with propositions, it should be clear that
situations, events, attributes, and utterances may be (and routinely
are) treated as evidence – or, perhaps more precisely, as the primary
sources from which evidence in the form of propositions may be
derived.41

Evidentiariness as a Matter of Degree

The usefulness of such a taxonomy is perhaps compromised, however,

by its ignoring the most essential aspect of evidence, which is its
usage by people in argumentation.
The general definition of evidence given earlier is in one significant
sense a simplification that serves to hide an important level of com-
plexity. Such complexity is a result of the participation of people in
the inference process – people whose prerogative it is to decide, in any
given case, whether or not to infer a particular conclusion. Like all
propositions, conclusions are the objects of personal evaluation; they
are subject to belief or disbelief. It can be useful, therefore, to view
evidence more formally as that which serves as (or that which pro-
vides) grounds for a person’s belief in the truth of a target proposi-
tion or conclusion. We might wish to say, for instance, that a
proposition p is evidence for q if it is grounds (i.e., reason, cause,
warrant) for believing q. Alternatively, we might prefer to assert that
a belief in (the truth of) p is evidence for q if it warrants belief in (the
truth of) q. Of course, unless we are careful to describe the nature of
the kinds of grounds for belief that are meant here, this definition will
remain unhelpfully tautologous: we are simply substituting ‘‘grounds
for belief’’ for ‘‘evidence.’’ Adopting a pragmatic perspective, we may
recognize that p is grounds for belief in q, if a rational person’s

41
The act of inferential analysis, carried out with a view to deriving conclusions from premises,
should perhaps be conceived as distinct from the act of interpretation that is undertaken in order to
derive premises from primary sources. Nevertheless, it should also be clear that the results of
inductive inference are also strongly dependent on individuals’ personal interpretations of the
concepts and categories that are manipulated in argument.
256 JONATHAN FURNER

believing p increases the probability that that person will believe in q.

Moreover, we may recognize that the degree to which p is grounds
for belief in (or is evidence for) q varies with the extent to which a
rational person’s believing p increases the probability that that person
will believe in q. Evidentiariness, in this sense, is not a binary prop-
erty, but one whose values vary continuously. It is more useful to
view evidentiariness as a matter of degree, where it is possible to
quantify or evaluate the extent to which one proposition (or belief in
that proposition) warrants belief in another.
Noting the distinction drawn at a separate earlier point between
beliefs (attitudinal thought-instances) and propositions (thought-
types), we may now distinguish additionally between a ‘‘strong’’ and a
‘‘weak’’ reading of evidence as grounds for belief that cut across the
sub-categories of ‘‘Evidentiariness as a property of ideas’’ identified
above. Before we can discuss these readings, however, it is first neces-
sary to clarify what is meant by another term. ‘‘Rationality’’ (or
sometimes ‘‘reasonableness’’) is the common label for a quality that is
often attributed to an evaluator or judge who makes epistemic deci-
sions – decisions about the truth-value or epistemic status of proposi-
tions that translate into attitudes towards those propositions – on the
basis of what the evaluator judges to be sound or valid arguments
made up of true premises. In legal contexts, an assumption is typi-
cally made that it is possible to estimate the probability that a
rational person will evaluate some specified target proposition or con-
clusion as true, by considering the degree to which their attitudes to
the premises and supporting argument are positive.
On a ‘‘strong’’ reading of evidence as warrant for belief, then,
something is called ‘‘warrant’’ if it actually (not only potentially)
increases the probability that a rational person will evaluate some
specified target proposition as correspondent to reality or as having
some preferred epistemic status. The things that may serve as warrant
in this sense are not certain other propositions, but particular beliefs
in (positive evaluations of, preferential attitudes toward) certain other
propositions as true, and beliefs in (positive evaluations of, preferen-
tial attitudes toward) certain arguments as valid. In this sense, the
warrant for a belief in a proposition is made up of other beliefs.
On a ‘‘weak’’ reading, something is called ‘‘warrant’’ if it is a
proposition, belief in which would potentially increase the probabil-
ity that a rational person will believe in another proposition. In this
sense, the warrant for a belief in a proposition is made up of other
propositions.
 CONCEPTUAL ANALYSIS 257
A Taxonomy of Kinds of Inference

Having considered the kinds of thing that evidence can be, I shall
now turn to a presentation of a taxonomy of the kinds of conclusion
that may be drawn as a result of the interpretation of a primary
source. What kinds of conclusion or inferences may we potentially
draw (or wish to draw) from our analysis of any evidential premises
that may be derived from a primary source such as our inscribed
vase, beyond simple reports of our perception of its existence and
physical attributes? To answer this question, first it should be recog-
nized that the premises that we derive from the vase itself are far
from the only premises that may be used in deriving a conclusion
about the vase. We have access, of course, to whatever set of proposi-
tions (and whatever set of evaluations of the truth of those proposi-
tions) that have been amassed from previous interpretations of objects
similar (in various ways) to the vase under current examination.
1. Inferences to context
– Deductive inferences to particular context – about the circumstances
and/or conditions of this particular object’s creation, including (specifi-
cally) the identity and nature of its creator(s). For example, we might
know from prior interpretation of the available evidence that vases made
of a particular kind of stone were made in Ireland (Premise 1: p fi q);
observe that our vase is made of that kind of stone (Premise 2: p); and
deduce that our vase was made in Ireland (Conclusion: q).
– Inductive inferences to general context – about the circumstances and/or
conditions of the creation of other members of any class of which this
particular object is understood to be a member, including (specifically)
the identity and nature of their creator(s). For example, we might know
that another vase was made of the same kind of stone and that that vase
was made in Ireland (Premise 1: p  q); observe that our vase was made
of that kind of stone and know from prior interpretation of the avail-
able evidence that it was made in Ireland (Premise 2: p  q); and infer by
induction that vases made of that kind of stone were made in Ireland
(Conclusion: p fi q).
2. Inferences to function
– Inferences to intentional function

Deductive inferences to particular intentional function – about the functions
or uses to which the creator(s) intended to put this particular object, and thus
about the reasons the creator(s) had for creating it.

Inductive inferences to general intentional function – about the functions
or uses to which the creator(s) intended to put other members of any
258 JONATHAN FURNER

class of which this particular object is understood to be a member, and

thus about the reasons the creator(s) had for creating them.
– Inferences to conventional function

Deductive inferences to particular conventional function – about the
functions or uses to which this particular object has historically been
put.

Inductive inferences to general conventional function – about the func-
tions or uses to which other members of any class of which this particu-
lar object is understood to be a member have historically been put.
3. Inferences to meaning
– Inferences to intentional meanings – about the thoughts that the creator
intended to express through the symbolic content of the object.
– Inferences to conventional meanings – about the meanings that have
historically been assigned by readers to the symbolic content of the
object.

Documents and Information

In the final sections of this paper, the preceding discussion is used as a

basis for drawing two tentative conclusions. The first of these, on the
comparability of evidentiariness and relevance, involves a comparison
of the concerns of archival science and information science, and it may
be helpful if, in this and the following section, the nature of the rela-
tionship between those concerns is explored in a little further detail.
Archives work (and archival science) may be distinguished from
librarianship (and library and information science), and again from
museum work (and museum studies), in many respects. The histories
of the respective professions and associated disciplines have unfolded
separately over many centuries, and obvious points of difference
include goals, principles, and methods. It is also clear, however, that
the three communities share a concern with a common object of
attention – the human artifact.
Librarians and information scientists are concerned with the pres-
ervation and provision of access to artifacts that have symbolic con-
tent of the kind described earlier. It is in virtue of this content that
such artifacts are characterized as documents; it is this content – or its
container or its meaning – that is commonly identified as information.
For librarians, the most interesting property of artifacts is their
‘‘documentness’’ – i.e., their informativeness, the degree to which they
are informative, their status as information.
 CONCEPTUAL ANALYSIS 259
Librarians’ conceptions of the way in which documents are infor-
mative have varied historically.42 At the objectivist end of the spec-
trum of conceptions, documents are treated as containers of meaning
whose movement from one location to another wholly constitutes the
process we know as communication. According to this reading, the
informativeness of a document is an inherent property of that docu-
ment. At the subjectivist end of the spectrum, documents are treated
as objects to which meaning is assigned by the reader, and informa-
tiveness is assumed to be a property of the document–reader pair. By
the objectivist account, it is possible to determine the degree to which
a document is informative by calculating the probability of its being
selected by chance from the universe of potential documents. Subjec-
tivist theories suggest that informativeness is more properly equated
with relevance, and measured by calculating the probability with
which a given reader will express a preference for a given document.
Notwithstanding this variation among conceptions of informative-
ness, at all points of the spectrum a distinction is commonly made
between two essential components of documents: the container, sig-
nal, or utterance on the one hand, and the content, message, or
thought on the other. The relationship between signal and message –
between utterance and thought – is one of meaning; a document, as a
symbolic entity, is meaningful to the extent that it expresses thoughts,
or mental entities. In turn, thoughts may themselves be conceived as
having content, in the form of the physical situations or events (actual
or potential) that are the objects of those thoughts.

Records and Evidence

Documents and other artifacts stored in archives are often considered

as potential evidence (or potential sources of evidence) of events that
occurred in the past. Such artifacts are known as records in virtue of
this potentiality: an artifact that may potentially serve as the source
of grounds for believing that a particular event happened is a record.
For archivists, the ‘‘recordness’’ of an artifact – its status as a
record – is its most significant property. An artifact is a record to the
extent that it is potentially evidentiary. Another way of talking about
the recordness of an artifact, then, is to talk about its evidentiariness.

42
See, e.g., Rafael Capurro and Birger Hjørland, ‘‘The Concept of Information’’, Annual Review
of Information Science and Technology 37 (2003): 343–411.
260 JONATHAN FURNER

P1 R asserted that document X2 is authentic, and p  q.

R believed that document X2 is authentic.
P2 R asserted that document X3 is authentic, and p  q.
R believed that document X3 is authentic.
Pn R asserted that document Xn is authentic, and p  q.
R believed that document Xn is authentic.

C1 / P3 If R asserted that document Xi is authentic, then p fi q.

R believed that document Xi is authentic.
P4 R asserted that document X1 is authentic. p.

C2 R believed that document X1 is authentic. q.

Suppose that we have before us a record – one that appears to be a

report of the minutes of a meeting of the members of a particular
organization. I say ‘‘appears to be’’ since, as well as representing in
writing the content of the discussion that supposedly went on at the
meeting, the record contains a certain amount of metadata about
itself, including an indication of the date on which it was authored
(not necessarily the same date as that on which the meeting took
place), a note of the identity of the author, and a certification of
authenticity – a statement, assertion or guarantee made by the author,
or by a third party, that may take the form of a simple signature – by
means of which it is certified that the document is, in actual fact,
what it purports to be. What is the process of reasoning through
which we may arrive at the conclusion that the meeting reported in
this document actually took place?
First, we must establish the reliability of the authenticator, R. If
that person were demonstrated to have behaved unreliably in the
past – if it were discovered that the authenticator sometimes fails to
faithfully express their actual beliefs about the authenticity of docu-
ments – then we would be forced to conclude that there is some reason
to believe that the authenticator did not believe in the authenticity of
the document in question, that this document is in fact inauthentic,
and that the meeting reported in the document did not take place. So
we must first consider what we might call the evidence of reliability.
Second, we must establish the authenticity of other documents
similar to this one in some relevant respect. If these other documents
were demonstrated to be inauthentic – if it were discovered that the
certifications of authenticity of these documents sometimes fail to
correspond accurately with their actual authenticity – then again we
 CONCEPTUAL ANALYSIS 261

P1 R believed that document X2 is authentic, and p  q.

document X2 is authentic.
P2 R believed that document X3 is authentic, and p  q.
document X3 is authentic.
Pn R believed that document Xn is authentic, and p  q.
document Xn is authentic.

C1 / P3 If R believed that document Xi is authentic, then p fi q.

document Xi is authentic.
P4 R believed that document X1 is authentic. p.

C2 Document X1 is authentic. q.

P1 If authentic document Xi exists, then p fi q.

event Yi occurred.
P2 Authentic document X1 exists. p.

C1 Event Y1 occurred. q.

would be forced to conclude that there is some reason to believe that

the document in question is in fact inauthentic, and that the meeting
reported in the document did not take place. So we must consider
what we might call the evidence of authenticity.
The proposition that document Xi is authentic is equivalent to the
proposition that the occurrence of event Yi (i.e., the event that is spec-
ified by the metadata associated with document Xi) is a necessary
condition for the existence of document Xi. In other words, if Xi
exists, then Yi occurred. Thus, a simple deduction on the basis of
what we might call evidence of activity is all that is needed to establish
the truth of the belief that event Y1 occurred.

Evidentiariness as Relevance

Distinctive conceptions of information abound in the literature.43 It

remains difficult (and it would not necessarily be desirable) to identify
a single conception that is useful in all or even most circumstances,

43
See Jonathan Furner, ‘‘Information Studies without Information’’, Library Trends 52(3)
(2004): 427–446.
262 JONATHAN FURNER

even within the relatively narrow scope of the library and information
sciences. The same can be said of conceptions of evidence in the con-
text of the archival sciences.
It is possible, nevertheless, to compare individual pairs of concep-
tions of evidence and information with a view to establishing the
extent and nature of any overlap. One of the more promising approa-
ches might be to focus on the shared propositional nature of two
conceptions of relatively wide acceptance and application, viz., docu-
ments as the sources of potentially relevant ideas (i.e., information),
and records as the sources of potentially evidentiary ideas. Taking
this approach might have the interesting effect of revising the motiva-
tion for the initial comparison, so that the objective is recast as one
of relating not evidence to information, but evidentiariness to
relevance. Just as evidentiariness is a property of the relationship
between one set of propositions (i.e., premises) to another (a conclu-
sion), relevance is a property of the relationship between one set of
propositions (i.e., a document) to another (a query). Sets of premises
are evaluated in order to measure the weight of evidence they supply
for conclusions; documents are evaluated in order to measure the
degree of relevance they have to queries. Both evidentiariness and rel-
evance are matters of degree; both are probabilistic. On this reading,
to seek to define the link between evidence and information is per-
haps to set oneself the less interesting challenge.

Archival Science as Social Epistemology

The second conclusion to be drawn in this paper is the result of a com-

parison of the goals of archival science with those of a particular sub-
field of contemporary epistemology. Epistemology may be considered
as a field of normative inquiry to the extent that its goals are assumed
to include (i) to establish a method of determining the likelihood with
which our taking a particular course of action will allow us successfully
to acquire knowledge in the future, consequently (ii) to enable us to
rank, in order of expected success (i.e., expected epistemic value), those
courses of action that are candidates in any given situation, and thus
(iii) to guide our knowledge-seeking practices. In these terms, the cen-
tral question that epistemology poses is not to be construed as ‘‘How
do we acquire knowledge?’’ or even as ‘‘How have we gone about suc-
cessfully acquiring knowledge in the past?’’ but ‘‘How ought we to go
about acquiring knowledge, given our particular interests and priori-
ties, and given the knowledge that we already have about the world?’’
 CONCEPTUAL ANALYSIS 263
On this reading, epistemology may also be treated as a naturalistic
discipline to the extent that it recognizes the possibility of meeting
these normative goals only once account is taken of empirical data
about the natural world – specifically, the actual epistemic values
assessed of past courses of action. A naturalized epistemology is one
that itself relies on the distinctive knowledge-generating method of the
natural sciences to provide support for its normative claims, rather
than seeking confirmation purely through rational argument.44
A veritistic epistemology is one in which any given belief must be
true before it may count as knowledge; a reliabilist epistemology is one
in which, furthermore, the status of any given belief as knowledge
depends on its having been generated as a result of a procedure that
has produced mainly true belief in the past – i.e., a procedure that is
generally reliable in its propensity for causing belief that is true rather
than false. Epistemology may further be characterized as social to the
extent that its assessments of the epistemic value of past and future
courses of action require prior evaluation of the activities of parties
other than the individual knower. Many of the personal beliefs held by
individual knowers are caused by those people’s decisions to accept as
true, and to treat as evidence in support of further conclusions, the
statements made by others. People who are evaluated as reliably truth-
telling are said to be trustworthy; their past reliability justifies our
continuing to place our trust in them, and to accept their testimony as
true without seeking additional corroboration. The normative goal of
one dominant variety of social epistemology – a veritistic flavor pri-
marily developed by the reliabilist Goldman – is to specify the kinds of
social practice that we may rely on most surely to generate true
beliefs.45
The program outlined by Goldman has found application in juris-
prudence and in information science, among other cognate fields. In
both cases, the application predates its being labeled explicitly as
‘‘social epistemology.’’ Focusing on the legal regulations that specify
the kinds of evidence that are admissible in court, Leiter46 suggests
how Goldman’s framework may inform an existing strand of legal
scholarship that seeks to measure the effect of individual rules of
44
See, e.g., Philip Kitcher, ‘‘The Naturalists Return’’, Philosophical Review 101(1) (1992):
53–114.
45
See Alvin Goldman, Knowledge in a Social World (Oxford, 1999); and Goldman, ‘‘The Need
for Social Epistemology’’, in Brian Leiter (ed.), The Future for Philosophy (Oxford, 2004),
pp. 182–207.
46
See Ronald J. Allen and Brian Leiter, ‘‘Naturalized Epistemology and the Law of Evidence’’,
Virginia Law Review 87 (2001): 1491–1550; and Leiter, ‘‘Prospects and Problems for the Social
Epistemology of Evidence Law’’, Philosophical Topics 29(1/2) (2001): 319–332.
264 JONATHAN FURNER

evidence on the likelihood that jurors will draw true conclusions

from the evidence presented to them. Fallis47 describes, in social-
epistemological terms, how readers of web pages can help themselves
in their searches for true information if they succeed in taking
account of contextual indicators of the veracity of individual pages’
content.
Given such models, it is possible to imagine how archival science
might adopt the framework of veritistic social epistemology with a
view (i) to improving our understanding of the nature and compara-
tive significance of attempts to clarify the concept of archival evi-
dence, and (ii) to establishing heuristics for selecting among candidate
procedures for obtaining knowledge from archival records. Those
models of archival practice that require the authenticity of the record
and the reliability of its creator to be established before its evidential
value can be assessed appear to share the reliabilist’s distinctive
assumption that knowledge is belief caused by a generally reliable
process, and it would no doubt be of interest (both to epistemologists
and to archival scientists) to investigate further the value of social
epistemology as a theoretical foundation for archival science.

Conceptual Analysis as a Method of Promise

As was stated at the outset, the aim of the present paper has been
to assess the utility of a particular method – namely, conceptual
analysis – when applied in addressing a particular problem – that of
improving understanding of the nature, scope, and function of the
phenomena that are commonly grouped by archival scientists under
the heading of ‘‘evidence.’’ It was assumed that, if such understanding
were to turn out to be enhanced as a result of the application of such
a method, then not only would we be enabled to generate interesting
ideas for future research, but we would also thereby be justified in
evaluating that application as successful, and in evaluating the
method as potentially of more-general utility in archival science.
In the paper, the meaning of the concept labelled ‘‘evidence’’ was
defined as a class of phenomena – viz., those things that we consider in
order to infer a conclusion. Several candidates for necessary conditions
of evidentiariness – relationality, measurability, physicality, substanti-
ality, meaningfulness, objectivity – were identified and examined, and
47
See Don Fallis, ‘‘Veritistic Social Epistemology and Information Science’’, Social Episte-
mology 14 (2000): 305–316; and Fallis, ‘‘Social Epistemology and Information Science’’, Annual
Review of Information Science and Technology 40 (forthcoming).
 CONCEPTUAL ANALYSIS 265
taxonomies were presented (i) of multiple kinds of evidentiariness,
varying in the appropriateness with which they are attributable to situ-
ations, events, utterances, and ideas, and (ii) of multiple kinds of infer-
ence, varying in their purpose, in their level of generality, and in the
extent to which they capture intentionality and conventionality. On
the basis of this conceptual framework, a correspondence was shown
to exist between the use of the concept of evidentiariness in archival
science and the use of the concept of relevance in information science;
and an analogy was drawn between the purpose of archival science
and that of social epistemology. It is asserted that these correspon-
dences are worthy of further investigation, and that their identification
allows for innovative thinking about the nature and purpose of archi-
val processes and the function and design of archival systems. On the
basis of the evidence thereby provided, we may tentatively infer that
conceptual analysis may profitably be used to improve understanding
not just of the concept of evidence (central as it is to archivists’ con-
cerns), but of other concepts germane to archival science.

View publication stats