Problems with Provenance

Chris Hurley has a B.A. (Hons) in History (U. of Sydney) and post-graduate diplomas in Education (U. of Sydney), Librarianship (U. of N.S.W.), and Archives (U. of London). He began working in archives in 1971 at the Australian Archives where he was involved at various times in arrangement and description, transfer and disposal, and the development of the 1983 Archives Act. From 1981 to 1990, he was Keeper of Public Records for the State of Victoria. He has served four terms on the Council of the Australian Society of Archivists (ASA) and has convened both the A.C.T. and Victorian Branches. He is currently Convener of the ASA's Education Committee.

Abstract

To many archivists, documenting provenance means identifying who created a body of records which is being described. This has traditionally been done by nominating a single person or corporation as records-creator. Changing administrative patterns and recordkeeping practices make it increasingly difficult and inappropriate for ideas about provenance to be limited by the need to establish identical boundaries between records-creator and records created. The provenance of records is established by showing a relationship to (rather than an identity with) their context. Once free of the self-imposed obligation to make records and provenance statements co-extensive, archivists can explore new and valuable ways of enriching provenance statements by documenting many relationships between records and a multiplicity of contextual ideas.

Problems with Provenance

By Chris Hurley

Archival theory identifies two kinds of records-creators : human and corporate1. The process of records-creation is seen as fundamental to archival descriptive theory. In this and a companion article (in which I will deal with the value of statements of functional responsibility in delineating context), I wish to explore some facets of the idea of provenance. In particular, I ask two questions -

• How adequately do archivists define (or identify) the human and corporate agencies whose records-creating activity they see as fundamental to that idea?
• How adequately do archivists' ideas about what constitutes records-creation serve their needs when documenting recordkeeping activity?

An examination of the literature reveals that all too often these agents of records-creation (which are major conceptualisations within our descriptive theory) are not defined, ill-defined, or defined exclusively in terms of the records-creation process itself. Thus, the ICA's standard for archival description emphasises the close affinity between the definitions of records-creator and the records created -

Corporate body. An organization or group of persons that is identified by a particular name and that acts, or may act, as an entity.

Fonds. The whole of the documents, regardless of form or medium, organically created and/or accumulated and used by a particular person, family, or corporate body in the course of that creator's activities.

Provenance. The organization or individual that created, accumulated and/or maintained and used documents in the conduct of personal or corporate activity.2

Although `corporate body' is here defined independently of ideas about records-creation, it is clear that its only use is to serve as counterpoint to `person' in defining the `fonds'. Once that is achieved a familiar circular process of definition occurs -

Question : Who created this fonds?
Answer : The records-creator (provenance) created this fonds.
Question : Who is the records-creator (provenance)?
Answer : The creator of this fonds is the records-creator (provenance).

In the case of natural persons, we assume there is no difficulty. The ICA standard does not even bother with a definition of `individual' (person). When describing personal papers, archivists do not combine two natural persons into one human records-creator (or sub-divide one person into several records-creating parts). They may be unclear about identification - whether they are dealing with one person or two, whether or not two names apply to one individual - but by and large they are comfortable with the idea of independent person-hood.

When documenting personal papers, establishing who created the records thus adds value to the description because, even in archival descriptive practice, persons are understood to have identity and definition apart from their records. Only the most hardline recordkeeper would insist on defining homo sapiens as a `records-creating mammal', but something very like this occurs when we come to deal with corporate records.

It is the thesis of this study that the idea of `creatorship' is most central to the concept of the fonds. Individuals and institutions produce records naturally in the course of undertaking their normal functions and activities. Whether they actually originate the records, receive the records, or share and manipulate information that is in or could become records, they create an aggregate of documentary material, in whatever form or medium, that reflects their juridical status. The resulting `natural' or `organic' aggregation of records is called a fonds.3

In traditional descriptive theory, identification of the records created with the records-creator is fundamental because records (or sub-sets of them) are the sole object of description. More recently, growing numbers of archivists have perceived the advantages (for information management and retrieval) of separating the two, but I hope to show that this is not enough so long as ideas about records creation are still bound up in the separate descriptive entities thus identified. Traditionally, any separation between the boundary of the fonds (or recordkeeping system) and the definitional boundary of the records-creator confuses the provenance statement because it is fundamental to the method that the two should be seen as identical (`an aggregate of documentary material ... reflect[ing] their juridical status'). Australian practice4, which has long separated them for the purpose of showing two or more records-creators in succession to each other, might (though it has not) have allowed for a separation also between the definitional boundary of the records-creator and the body of records created.

It is a curious irony (in view of what I am about to write) that one of the things most consistently charged against the Australian system has been an alleged failure to respect des fonds (as it were) by permitting a confusion of the provenance statement. This allegation has always been based on a misconception about Australian views concerning provenance, which have focussed (as I shall demonstrate) in a very traditional way on the single agent of creation - at any given time. In this article, I shall at last commit the heresy of which we have for so long stood accused, nailing this proposition to the door : that the provenance statement indeed can and should be confused (I would say clarified) by permitting simultaneous attribution to more than one creator. In anticipation of further transgression, I shall try to lay the groundwork for another idea : that provenance cannot be adequately described if limited to showing agents of creation.

Both traditions have tended to define corporation as a `records creating organisation'. Without more, such attribution is tautological and adds absolutely no value to the description of the records. When we say `this is the personal correspondence of the First Duke of Wellington', we are adding value to the description because of all the contextual information which attaches to the name `First Duke of Wellington'. If we say `these are the records of the creator of these records' we are saying virtually nothing. Value is added to such a description only because archivists insinuate into their description of provenance information about the records creating agent. Thus, a description of the Duke of Wellington's personal correspondence will contain much data about the Duke, and a description of `these records' will contain much information about `the creator of these records'.

Archivists have adopted rules to ensure that a single records entity is shown as resulting from the creative action of a single person or corporation because they have not bothered (?needed) to be careful about teasing out the different strands which make up the idea of provenance. Australians have challenged those rules by insisting that it is both desirable and legitimate to show that records can have multiple creators successively, but we too have adhered to the idea that there can only be a single records-creating agent at any given point of time.

When dealing with persons, this requires a conceptual separation of a natural person in his `private' capacity from any potentially conflicting capacity as an official or agent within a corporate or family grouping. All sorts of stratagems and devices have evolved (where there is not just downright confusion) to preserve the primacy of the single records-creator producing an `organic' or `independent' set of records. These are broadly of three kinds:

• A definite distinction may be made between personal and official records. The records from a Minister's office can be variously shown as part of his personal archive, the archive of the office he held (during his occupancy of it), or (more rarely) the archives of the ministerial portfolio which he held as one in a succession of office-holders.
• A rule of convenience may be established in each case to keep the identity between records and creator intact. When we wish to keep official estrays with personal papers, we stress the importance of preserving evidence of last practical use. When we wish to restore them, we stress the importance of maintaining the records-creator's intention that they be part of the system to which they originally belonged.
• If the records-making of two or more natural persons is inextricably mixed, we go up one level and identify them as `family papers'.

The single-minded pursuit of this idea of provenance has one great advantage. Archivists have established a unified, coherent, standardised perception of what provenance means and sustained it over a long period of time. Like Darwinian theory, it is not perfect but it is better than the alternatives. Archivists needed, above all, to defend their methods against the threat of confusing records-maker with the authors or subjects of correspondence and, more significantly, against those who would disturb the records according to such information-based ideas. Once multiple-provenance is admitted, the basis for defending the archives against re-arrangement (maintaining its evidential qualities - its functionality, if you like) is weakened.

Figure One

It is arguable, however, that this justifiable defence of provenance has led archivists to an unduly narrow and increasingly inappropriate view of the matter. The First Duke of Wellington (viewed purely as records-creator and ignoring, for the moment, any distinction between the natural and official person) is going to provide a rich and complex provenance - but a few aspects of a long and varied career are set out in Figure One.

Whether archive groups or fonds corresponding to those I have identified in relation to Wellington actually exist or not, I have no idea. I may fairly suspect that many archivists will be found who dispute the legitimacy of some at least of the groupings I have identified - my separation of ministerial and departmental records, for example, or my treatment of parliamentary archives. It is central to my thesis that debate about the contours of middle level contextual entities of this kind is vital to the development of more sophisticated ideas about provenance. A debate on the merits of these particular examples is obstructive, however, of my purpose here which is to illustrate the need for such ideas. At the risk, therefore, of incurring criticism that the problems I now wish to explore arise only because of my own carelessness in formulating the hypothetical contextual entities I wish to discuss, I will not become side-tracked into a defence of the merits of the hypothetical groupings I have chosen.

It should be noted also that in the ensuing discussion I have deliberately limited myself to ideas about provenance as they relate to one level only of recordkeeping activity - viz. describing aggregations of records (whole series, recordkeeping systems, fonds, etc.). This is how archivists traditionally think of provenance statements. Recordkeeping activities (each involving ideas about provenance) are undertaken, however, at many levels - the raising of a single document, for example, within a transaction the record of which may itself be incorporated into a series, system, fonds, etc. Margaret Hedstrom has indicated for us the variety of levels at which archival data on provenance may be needed in the design and administration of electronic recordkeeping systems5.

Problems with provenance arise because it will not always (perhaps seldom) be appropriate to nominate as `creator' of a record (or of the documents which make it up) the corporation or person responsible for creating the records into which the record (including the documents which make it up) is aggregated for recorkeeping purposes - the most obvious example being the activity of records-creating individuals or divisions within a records-creating corporation. An exploration of problems with `multi-layered' provenance is beyond the scope and intent of this article (perhaps next time). If my thesis here is accepted (viz. that the provenance even of the whole recordkeeping entity cannot be adequately dealt with using a single idea), it should be but a small step to apply that insight to dealing with multi-layered provenance.

Both the natural and official persons, Arthur Wellesley, First Duke of Wellington, occupy an important place in delineating the provenance of several different record groupings but, with the definite exception of his own personal `private' papers, it is impossible to regard him as sole records-creator in most cases. At best, traditional theory can only be sustained by -

• dividing Arthur Wellesley into sub-sets representing his various official posts or functions (hats) - thus, Arthur Wellesley (Prime Minister, 1828-1830) might be regarded as a different records-creating person from Arthur Wellesley (British and Allied Commander, 1815), or
• submerging Arthur Wellesley altogether as a records-creating person in relation to official records and regarding him purely as an operative within the administrative or military units in which he functioned.

This cannot be right. To say that `these are the records of the British Prime Minister, who was then the Duke of Wellington' is to say more than `these are the records of the British Prime Minister in 1829'. Most Prime Ministers (and especially this Prime Minister) are not faceless functionaries; their personality and individual character are an important part of that knowledge about the context of records which the delineation of provenance provides.

Taken to its logical conclusion, this view annihilates the distinction between personal and official records : all records have both a corporate and a personal provenance. There is, however, a distinction to be made between Arthur Wellesley, the creator of his own personal papers, and Arthur Wellesley, whose activities are documented in the official records. Archival theory should allow for both these important facets of the provenance of the records concerned to be demonstrated without confusion wherever appropriate. Because it does not, because it deals only with the question `which single person or corporation created these records?', archivists must ignore (or at least downplay) important aspects of provenance.

Let me stress that I am not arguing in favour of identifying Wellington as the Prime Minister who created the official records because he was a notable figure. Indeed, I would argue that, as we explore the facets of multiple-provenance more deeply, the greatest danger to be avoided is any confusion between linkages (relationships) which are established to show provenance and others which are designed to retrieve on the basis of different ideas (e.g. subject). Wellington's Prime Ministerial records would, no doubt, contain information about numerous notable figures of his time - it is not therefore appropriate to include them in any statement of provenance. The argument is that knowledge of the personal identity of the records-creator in a case like this (bearing in mind that each person is potentially a records-creator in his own `private' capacity) is important (arguably as important as knowledge of the office he held) when describing their provenance, origin, or context - whatever idea you think it is that is being documented.

In fact, no one disputes this. Supposing a separate fonds for British Prime Ministers did exist, it is inconceivable that its description would not include an extensive account of the life-history and career of each of the incumbent Prime Ministers who held office during the date range covered by the fonds. Similarly, in the Australian system, any account of the Prime Ministership, developed independently of records descriptions and subsequently related to them, would dwell at length on an account of the persons who held that office. When identifying a contextual entity, let it be noted, we will fill pages with information about personality, activity, and function - yet (even though it is the chief defining characteristic we use to identify such entities) it will be a rare archivist indeed who gives records-making itself any prominent part in his account of a contextual entity.

The plain fact is, what everyone knows, that Arthur Wellesley played an important and varied role in the history of his country and, in consequence, created records in a variety of different capacities which are likely to be scattered throughout numerous fonds of which he may or may not be shown as records-creator according to traditional theories - depending on the primacy given by the descriptive archivist to that aspect of their provenance. Arthur Wellesley is no less the creator of the Prime Minister's records in 1829 than he is of his own personal papers in that year (though his role is clearly different in each case). Following their rule against simultaneous multiple provenance, most archivists will choose to recognise this when describing the personal papers and suppress it when describing the official records.

In short, archivists have an inadequate idea about provenance because -

• they cannot articulate the variety of different roles or relationships a corporation or person may have in relation to the records-creation process,
• they are bound by their theory to identify one only and ignore the rest, and
• their definition of that one is tautological.

When archivists identify the official prime ministerial records as the records of the Office of Prime Minister they are compelled by their own logic to say further that they are not the records of the records-creating individuals who occupied that office. Some archivists will say that I have misrepresented their position. They will argue, rightly, that information concerning the identity of the office-holder can be found in their contextual descriptions, that this information can be formatted into `access points' (as if information retrieval were the sole or even primary purpose of identifying provenance), and that the role of each individual in the records-creation process can, in consequence, be inferred - thereby adding to the explicit identification of the office which created the records an implied identification of each incumbent as `joint' creator. In the Australian system, the connection can be made another way, by documenting a relation between the person and the office so that the `joint' creation which must be otherwise inferred can be established systematically :

• Q. Who created the Prime Minister's records? A. The Prime Minister (agency 23).
• Q. Who was agency 23 in 1829? A. Arthur Wellesley (person 56).

On this view, Arthur Wellesley is conceptually divided for different purposes. The Arthur Wellesley who is depicted as the `sole' records-creator of his own personal papers is also shown as a `joint' records-creator indirectly by naming him as holder of the office which is nominated as the sole records-creator of his official records6.

The alternative view, more in tune with archivists' theoretical position, is to deny Arthur Wellesley (the person) any meaningful role when identifying the provenance of the 1829 Prime Minister's records. On this argument, the records are those of the Prime Minister, who happens to have been Arthur Wellesley but could just as easily have been Bugs Bunny. The identity of the individual who was the Prime Minister (however interesting that might be for information retrieval purposes) is irrelevant to its provenance - it is no more relevant than the identity of the file clerk who dockets a piece of incoming correspondence (one might say as relevant). This is not to promote ignorance of the personal identity of those involved in records-creation, it is rather to say that this information (however important in itself) must not be confused with provenance data which has the single explicit purpose of articulating who was responsible for making the records.

Can this single-minded view of the nature of records-creation process be sustained? I think not. Whatever view one takes of the personality of officials, there are unquestionably examples of family papers which result from records-creating activity undertaken jointly and not by single individuals. To maintain the fiction of the single, independent records-creator, recourse might be had (unworthily) to artifice. If, for example, letters to Arthur Wellesley and his wife Kitty were kept by them in a single chronological sequence, it might be argued that Arthur was the records-creator because he filed his wife's correspondence with his own. In some circumstances, this might even be the most accurate representation of what actually happened. The alternative possibility - that two related persons maintained their correspondence as one - is conceptually just as likely. If pressed to recognise this reality in his description, the archivist next has recourse to identifying the family of joint records-creators as the provenance -

... individual fonds are not always the records of a single creator empowered by a clearly defined set of functions and activities. Such non-corporate fonds may be family papers spanning several generations, the accumulations of a husband and wife or siblings jointly ... 7

This hardly removes the difficulty since the individuality of family members must still be recognised (unlike corporate entities where we can extinguish identity by merging or dividing units of contextual description according to our perception of their independence as records-creators, simply in order to suit our documentation procedures). Even this device will fail, however, when dealing with the records of a collaboration between unrelated individuals - if any archives survive of the partnership between Gilbert and Sullivan, for example.

It seems to me, therefore, that the possibility (conceptually at any rate) of simultaneous multiple provenance must, at least where natural persons are involved, be conceded - whatever stratagems may be used to disguise the fact when maintaining an untenable archival theory which denies that possibility. The last resort of an archival mind at the end of its tether might be to identify joint personal creators as a single entity - `How do you do. I am Sir Arthur Sullivan. He is Mr Gilbert. They are Gilbert & Sullivan'8.

The Australian descriptive practice of separating data on context from data on recordkeeping has led to surprisingly little original thinking on the nature of contextual entities. In particular, Australian ideas about provenance (records-creation) have remained at almost precisely the same level of development as those of traditional archives theory9. The principal Australian contribution to solving problems of simultaneous multiple provenance lies in the capacity which the Australian system has for linking another `creator' indirectly with records by establishing a contextual relationship with the `true creator' - vicarious simultaneous multiple provenance (!) as it were.

This sluggishness is not altogether surprising given the circumstances in which Australian theory was launched10. Urging `multiple-provenance' archiving to allow a single record-keeping entity to have successive creators and thus `belong', at different times, to successive fonds or record groups, each capable of being represented logically even though its physical features might no longer survive was (for its time) innovation enough. That it should still be attacked as heretical thirty years later suggests this caution was justified.

There was concern that, however much Australian practice might vary from the traditional, the departure should be seen as coming within the boundaries of accepted archival principles. The primary purpose of the innovation was to free documentation practices from the limitations imposed by focussing on a single phase in the record-keeping process. This involved allowing for successive phases in the record creating process to be shown by establishing relationships between records and successive records-creators. The key lay in allowing many to many relationships between records and records makers.

This innovation did not necessarily require any serious re-evaluation of existing notions of what records-creators were (though the failure to undertake such a re-evaluation may, with benefit of hindsight, be regarded as a missed opportunity). At any rate, Australian descriptive practice, despite many apparent differences with that elsewhere, has not developed any very distinctive ideas about the nature of contextual entities. There has been no serious Australian challenge to the notion of an identity between the records-creator and the records created. Multiple provenance has been allowed when describing the chronological dimension only (it may more properly be described as successive provenance). Simultaneous multiple-provenance (two or more corporations or agencies responsible for creating the same records or fonds at the same time) has been outlawed in Australia as vigorously as it would be, no doubt, by right-thinking archivists everywhere.

Thus Scott defined an `agency' (a records-creator) in very traditional terms -

An agency is part of an organization that has its own independent recordkeeping system ... it may be at any level in the administrative hierarchy ... Agencies are free elements; they can, and do, move from organization to organization with little or no change 11

in terms, moreover, which intentionally mirror his definition of `person' (a records-creator), described as an element to -

provide the necessary context for personal or family archives and for archives of an agency that may have been produced by a given person ... Persons are `free' elements, as they may pass from one family to another, females by marriage or adoption, males by adoption ... 12

It will be apparent that Scott's agency (despite superficial differences) is indistinguishable, in its essentials, from the definition of `provenance' given twenty five years later in the ICA standard (the identity of the agency and the fonds being an idea he repeated over and over in his articles in the 1970s and 1980s). This is an indication, I believe, of how well traditional archival principles of provenance were understood in what was then the Commonwealth Archives Office of Australia13 and how anxious Scott and his superiors were to be seen as operating strictly within those principles.

Scott's own account of this decision says as much -

Administrative change has also had its effect on our approach to the level of definition of `agency' or `record group' (the problem being similar whichever concept is used); that is, using the words of M. Duchein, there has in general been a shift from the `maximilist' position (`record group' as a whole ministry) to the `minimalist' (`record group' as a smaller sub-ordinate office) ... The relatively `maximalist' Jenkinsonian definitions of `Administration' and `Archive Group' were first considered ... as being co-extensive with a ministry or department of state, but by 1962 a decision had been made to establish `separate Archive Groups (rather than Sub-Groups) for substantially independent offices within departments ...

There is, of course, danger of moving too far in the `minimalist' direction. As Duchein points out, `one runs the risk of depriving the notion of `fonds' of all real meaning'. It is in part for this reason that the Australian Archives ... does not normally register as separate agencies the functional division, branches, or sections of an administrative entity; in doing so, we not only avoid the even greater instability of such internal divisions and sections, but also reflect the reality that such internal divisions do not, as a general rule, have separate general record-keeping systems of their own ...

Changes of name to agencies can cause debate as to whether or not one should recognize a new name as implying a new agency ... a change of name at the level of Commonwealth department of state (ministry) is recognized by the Australian Archives as resulting in a new agency, but minor variations in agency titles at lower levels may not be so regarded, unless there is a substantial accompanying [sic] change in function.

In departing from record groups or agencies co-extensive with a ministry/department of state, the Australian Archives has consciously recorded the relationship of subordinate agencies [associated with the department] ...

A new more precise definition of `fonds' or `agency' has also been needed to reflect this departure from the Jenkinsonian `Administration' ... In 1966, the decision was made by Dr K. Penny and P.J. Scott to define `agency' not so much by reference to its administrative status or administrative independence but principally by reference to its degree of independent record-keeping ... In this, we followed comments by Muller, Feith and Fruin, on what constituted a separate `fonds' ...

There are, consequently, cases of administrative bodies, with distinct titles, even established by statute, which are not treated as separate agencies, when they do not have independent records but use the filing system of an agency with which they are closely linked ...

In Canada, the criterion of independent record-keeping has now [1980] also been included in the definition of record group ...14

There is a figure of speech called metonymy, one example of which involves using the container for the thing contained - `The kettle boils'. James Thurber recounts how, when he was a child, he annoyed his teacher by discovering an inversion of this (using the thing contained for the container) - `Get away from me, or I'll hit you with the milk!' (this from a time when milk still came in bottles)15 . Archivists may be said to be using metonymy or else inverting it in much of their thinking about provenance. Either they use the provenance (container) to define the boundaries and identity of the records (thing contained) -

a fonds is an organic whole and ... any Administration, or one or more of its fonctionnaires, can create a fonds d'archives provided that these include résolutions or procés-verbaux; the inclusion of Archives of such a type making it autonome. Roughly speaking, we may take it that ... the qualities of a fonds d'archives depend on its including those which, when the administration which created it was active, constituted the final authority for executive action. For our purposes we may do better perhaps to represent this quality in terms of Administration rather than terms of documents, the forms of which, as we shall see later, are not necessarily constant. Fonds we may render, for lack of a better translation, Archive Group, and define this as the Archives resulting from the work of an Administration which was an organic whole, complete in itself, capable of dealing independently, without any added or external authority, with every side of any business which could normally be presented to it. 16

or they use the records (thing contained) to define the provenance (container) which occurs wherever a person or corporation is defined as a records-creating entity. They do this because they employ a very clumsy idea of what constitutes provenance17.

Effectively, archivists have documented provenance by seeking out an `independent', `organic' or `autonomous' descriptive entity (either the records or the records-creator) and assuming a one-to-one relationship -

... The records being described should represent a distinct and coherent whole, one that will illuminate, and not obscure, the context of activities out of which the records were created and maintained during their active life ... Applying the first degree of respect des fonds, (maintaining the whole of the records created and/or accumulated and used by a person or corporate body) requires the isolation and circumscription of the fonds creating body. The task is not an easy one given the enormity and complexity of the administrative environment in which records are typically generated. Where in the frequently bewildering hierarchy of records creators do we locate the fonds, the unbreakable whole?18

Either an autonomous records-creator is assumed to exist on the evidence of an autonomous fonds (Scott) or the fonds is defined as an independent, organic whole because it emanates from a functionally autonomous provenance (Jenkinson). Define them in any other way and it is at least logically possible that some records will be perceived to be the product of activity undertaken jointly by two or more persons and/or corporations.

Australian analysis of context today varies little from that given diagrammatically by Scott in 196619 and reproduced here (with some enhancements) as Figure Two. This analysis shows two levels of contextuality : provenance or records-creation (which will be easily recognisable to archivists everywhere) and `higher' level groupings or families of records-creators (which I have elsewhere20 described as ambient entities). At each level (provenance and ambience), there are parallel ideas dealing with corporate and human entities (organisations are the corporate equivalents of families and persons are the human equivalents of agencies).

In fact, very little work has actually been done on the human side of this model. Only in Australian Archives (so far as I am aware) has any extensive work been undertaken and, even there, families are no longer described and relatively few persons documented. Had the human side of the model been more extensively and fully developed, I believe the groaning inadequacies of the model itself would have become much more quickly apparent.

Figure Two

It is obvious that the parallel between human entities and corporate entities is strained and illogical. In Scott's model, an ambient entity is not (cannot) ever be shown as a records-creator - except indirectly through the provenance entities of which it is composed. Thus, an organisation is a construct or grouping of records-creating agencies, never a records-creator. By extension a family must be regarded in the same light. Yet the notion of family papers (as distinct from records created by individual members of a family) is common and well-accepted within the documentation programmes which deal extensively with records created by human entities and, as we have seen, may be the only way of maintaining the fiction that records are never the product of a process of joint records-creation. Leaving aside for a moment which of these approaches is preferable, what is plain is that the model cannot accommodate both.

Similarly, the notion of agency cannot exactly parallel that of person. An agency is in fact a unit only to the extent it is conceived of as such. An agency may itself be composed of sub-agencies or be itself a sub-agency within another. Conceptually, we may move without difficulty from a `maximalist' to a `minimalist' view or take up any other posture we like. Agencies exist, conceptually at any rate, as divisible portions of a divisible whole (just as a fonds can be divided into a sub-fonds, sub-sub-fonds, etc.). The idea that agencies exactly parallel persons (who are manifestly indivisible in any relevant way21 - the left elbow is never going to set up record-keeping on its own) but that both are conceptually distinct from the organisational and familial structures of which they compose a part is clearly inadmissible.

An ambient entity should move around and through provenance - like functions. Scott used organisations and families to combine agencies and persons together in larger groupings and forbad their use as records-creators, but in certain circumstances, as I hope to show (and as others have already realised), both organisations and families can be used as records-creators. More to the point, it is clear that Scott used agencies to show both provenance and ambience. He identified a superior/subordinate relationship between agencies which is indistinguishable from the relationship which subsists between agencies and organisations22. In Scott's system, a superior agency does not operate as a records-creator; its only purpose is to embrace or gather together one or more other agencies into a conceptual grouping (just like an organisation). Agency, as defined by Scott, is not used for provenance or ambience - it is used for both.

This is a good example of why it is unwise to define entities in terms of the use to which they will be put in documentation. An agency is not a `records-creator'. An agency is a corporate entity which may be put to any one of several possible uses - including documentation of records-creation. Neither it nor any other descriptive entity should be defined in terms of the descriptive purpose(s) for which it will be used because that information (information about context, provenance, and recordkeeping activity) is itself wrapped up in the relationship which is yet to be established. The use to which an entity will be put must be assumed if it is defined in those terms, yet it is the discovery of possible uses which is the very purpose of documentation activity. If we build assumptions about the thing we hope to discover into the conceptual tools we intend to use, those tools will be flawed. In a companion piece of work, I hope to demonstrate that functions can be usefully employed in crafting ambient relationships with corporations, but I would not wish to exclude the possibility of using them to show provenance by establishing relationships directly between functions and records.

I believe that both the fonds-based and (in mistaken imitation) the Australian descriptive traditions have failed to document provenance well because they have not developed an appropriate (or even adequate) theoretical basis for describing records-creating entities or articulating the variety of records-creating relationships which may subsist between them. Both traditions have been guilty of establishing a false parallel between corporate entities and persons and, in despair of an independently meaningful concept, both have fallen back on an essentially circular and meaningless definition of records-creator, the use of which, when establishing a relationship between the records-creator and the records created, adds little value to the description.

Bearman and Lytle23 suggest that this is because archivists have been too strongly influenced by what they term 19th century Weberian `classical organisation theory' which pictures corporations as simple, autonomous, and hierarchical. I think they attribute too much sophistication to the archival mind. Archivists, like almost everybody else, simply confused corporate agencies with persons.

I am not sufficiently well versed in the history of institutions to say confidently why this might be so. An explanation might be found in the observation that in the interval between Roman times and the nineteenth century, European institutions were small, simple, official rather than corporate, and perceived in personal terms24 -

What distinguishes government from personal control is its unremitting character. To be governed is to be subjected to the regular pressure of an authority operating according to fixed rules. In the full sense of the word, it is arguable that nobody was governed before the later nineteenth century; it would certainly be foolish to maintain that either royal or princely government in the twelfth century operated according to fixed rules or without intermission or over all the inhabitants of a defined area. Nevertheless, the perception that this might be desirable was reborn.

What provoked it was in part the ecclesiastical reform movement. For the reformers' vision of a fully-developed hierarchy, each man with his own sphere of competence ... evoked a mixture of admiration and envy in those laymen, accustomed to deference, whose position in the ecclesiastical scheme was lower than they would have wished ...

In returning to the old imagery of the prince as the bearer of the sword, the reformers had in fact triggered off an important set of associations. They recalled the Augustinian view of political authority as established not merely for defence against external enemies, but also for internal peace ...

Princes were happy to exploit those elements in the intellectual ferment of the twelfth century renaissance, which exalted their position as executors of justice ... The study of the classics, even when unfortified by specific reference to Roman law, was a potent weapon in strengthening princely jurisdiction at the expense of others ... So behind and above the person of the prince, there formed again the abstract concept of the `public person', the punisher of the wrongs and injuries, the bearer of peace, the mighty distributor of justice and equity ...

The intellectuals' contribution to the practice of government was, at least in the short term, less of a mixed blessing than their political and legal theorizing. It lay in producing those written documents - charters, writs, letters, accounts - by which hitherto spasmodic personal contacts between princes and their officials and subjects could be regularized ... The evolution of a rudimentary bureaucracy was, by 1180, the distinguishing feature of royal and princely administrators ...

The most famous chapter in the Policraticus [by John of Salisbury] is that which sets forth the organic image of the commonwealth, the prince the head, the church the soul, the senate the heart, the soldiers the arms, the peasants the legs - a classical cliché brought up-to-date. The importance of this image in the history of western political thought has long been recognised; indeed it has been hailed as the rebirth of the concept of state in the Middle Ages. Its most striking feature is the importance attributed to princely officials: judges and provincial governors are the eyes, ears and tongue of the body politic; officials are the hands; and financial officers the stomach and intestines. By the middle of the twelfth century, John's French readers would have had no difficulty in making the necessary identifications : castellans and viscounts, baillis and prévôts, household officers, the clerks, knights, and chamberlains of princely courts abounded. Bureaucracy, long absent from the country, was making a rapid return ... 25

Perhaps, too, we have been over-much influenced by the library cataloguing tradition which emphasises similarities between personal and corporate `authors' and treats problems of corporate identity principally in terms of change of name.

A generalisation comprehending the development of institutions across all the nations of Europe (including governmental, private, local, and ecclesiastical bodies) during one and a half millennia might give even me pause. I will venture to suggest, however, that, for a sufficiently long period to be significant, two factors (at least) bolster the conceptual parallel between persons and government institutions -

• Many European institutions were extensions of an office held by a person carrying out the duties of that office - the identity of the office (or department) derived from the functions of the office-holder whose activities it supported and little else, and
• The identity of the state itself was not clear - local loyalties, the absence of an abstract conception of nationhood, and the primacy of notions rooted in kingship and tribal peoples prevailed.

A medieval kingdom, for example, is better understood as a fluctuating geographical area (before that a tribal people) coming within the jurisdiction of an official (the king) - not, as we might understand it, a territorially defined political entity engaging its citizens in a mutual bond of rights and obligations26. A feudal hierarchy, though capable of sustaining complex overlapping jurisdictional responsibilities, was based on a personal allegiance to the sovereign27. The `state' hardly existed independently of the king28. What we would think of as executive departments were, in fact, extensions of the offices held by the king's servants (his officials) with their associated attendants29.

Formal ideas of corporate identity independent of the natural person developed slowly 30. The development of recordkeeping in close association with more complex corporate structures apparently makes a definition of corporations based on recordkeeping activity plausible31, but neither recordkeeping nor organisational complexity are things which are pursued for their own sake. It is clear that complex organisational structures were called forth only by a strong social and cultural forces and that the first stage (at least in the European experience) of developing complex government structures was the concentration of power into the hands of rulers and their officials and deputies32.

It is only in the last two hundred years that government corporations as we have understood them until recently - large, compartmentalised, hierarchically structured bodies - have developed. Even in what was once referred to as the `early modern' period, successful corporate structures responsible for the execution of policy were patterned upon the assignment of functional responsibility to officials (the diffuse Habsburg monarchies, it could be argued, failed because their extent and diversity demanded a complexity and size of administrative organisation which the society of the time could not sustain)33. In the private sector (at any rate under British law) legal incorporation on any widespread scale dates from only the nineteenth century under statutory schemes for the limitation of liability.

As late as 1851, the infant colony of Victoria provides an example at the frontier of European civilisation where the administrative arrangement of departments (immediately following separation from New South Wales) was based entirely on the division of responsibilities amongst colonial officials, formerly the Governor's servants, which is conceptually indistinguishable from those at the court of the Anglo-Saxon kings. The first Victorian Ministry comprised -

• Chief Secretary (formerly Colonial Secretary or CEO of the colony),
• Attorney-General (law)
• Treasurer (finance)
• Commissioner of Public Works (works)
• Commissioner of Trade and Customs (immigration and customs)
• Surveyor-General (lands)
• Solicitor-General (justice)

The same period which saw the development of modern, complex, departmental structures coincided, of course, with the development of archival theory. But archival theory had regard to the documentary survival of an earlier age - an age in which the parallel between persons and offices was strong. Archival theory was perhaps beguiled into thinking that records-creating corporations continued to be really very like records-creating persons with a clear-cut autonomous identity and to embody conclusions about these perceived similarities into its practice.

When corporations manifestly stopped behaving like persons - as they have for about two centuries - archivists failed to take stock. They went on treating corporations which were now behaving in a clearly un-human-like way as if nothing had happened. A succession of patched up solutions has been tried to overcome the theoretical problems created for us by this phenomenon. On this cosmic view of the matter, the Australian descriptive tradition may be regarded as just one more patched-up solution which overcomes the consequences of corporate record-creators behaving so unreasonably as to change their boundaries and identity frequently but it still fails to come to grips with the problem that organisational boundaries are not always the same as those of recordkeeping systems.

The chief consequence of this inadequacy has been to drive both traditions into a definition of provenance which is almost bereft of specific meaning. The practical implications of becoming trapped in this theoretical cul de sac are not without significance because we have entered an era in which the behaviour of corporations is undergoing yet another shift and a continued failure by archivists to develop a satisfactory view of corporations will (considering the centrality it has within our descriptive practice) have serious consequences.

Modern departmental structures (complex, truly corporate, compartmentalised, and highly structured) can exist where three pre-conditions have been met :

• there must be an identifiable notion of the state or patria which exists as a superior entity,
• the administrative need must be extensive, requiring a large and complex bureaucracy, and
• the social structure must be capable of supporting truly corporate (as distinct from office-based) administrative units.

Rather than insist on a manifestly inadequate conception of corporation, based on crude ideas (which do not meet our own needs, let alone anybody else's) about a corporation being an independent recordkeeping system, we would do better to recognise that corporations and ideas about them are related to fluctuating social and cultural conditions which affect both the changing nature of corporations themselves and our changing perceptions of them.

As corporations now become flatter and less compartmentalised, as workgroups replace traditional formal structures and networking breaks down the physical and work-process barriers between organisational units, the crude model that archivists use when designating provenance will reflect reality less and less. It will become harder and harder to maintain the fiction that corporate entities behave in an analogous way to natural persons and that they can be appropriately defined exclusively in terms of their activity as records-creators. Yet it is upon this fiction - that there continue to be independent, homogeneous, records-creating corporations from which there will emanate records constituting an identifiable, describable, independent `whole' - that archivists' central idea (almost their only idea) about provenance rests.

Terry Cook has identified a Canadian example of the problem (the Department of the Environment) in which departmental-wide concerns use one block of numbers in a file series while powerful sub-units (some of which pre-existed the department's formation) used discrete blocks of numbers of their own. He continues -

There are also the issues of simultaneous creators and of records-creating patterns in modern bureaucracies that do not conform to traditional, hierarchical, organizational theory. Indeed, some critics assert that the assignment of such series of records to any single, larger, "creator" entity is simply impossible given the state of modern administration and modern records-keeping.34

The concept of a single agency or fonds worked (more or less) at a practical level when archival description was focussed primarily on `holdings' and (to a lesser extent) when it focussed on manual systems managed by agencies which were, by and large, housed separately, each with its own defined boundaries and separate responsibilities, each with a structured hierarchy and its own unlinked recordkeeping activities between which communications definitely passed - in an environment, in other words, of separate work units where the communications systems and recordkeeping systems were also separate.

Suppose that the Government of Titipu comprises 22 departments, each with its own recordkeeping system - each maintaining adequate records of inwards and outwards communications. It decides to establish a wide area network (WAN) to link all 22 departments. Each department continues to maintain records of its communications (both internal and external). Then it is modified. Instead of each department keeping its own copies of inter-agency communications (resulting in two copies of each), it is decided to install a 23rd recordkeeping system to capture all inter-agency communications (together with proof of despatch and receipt and appropriate linkages back into the departmental recordkeeping systems). Each agency now has to create records of intra-agency and external communications only.

There are now 23 recordkeeping systems. What is the provenance of the 23rd system? There are only two choices. Either you must allow that the records are being created jointly by all 22 departments or you must 'go one level up' and identify the Government of Titipu itself as the records-creator. Circumstances might allow you to identify an independent agency which manages the 23rd system or to pretend that one of the 22 has `primary' responsibility, but this is no answer to the conceptual problem. Such a system, because there is only one `copy' of each message (even though dispatch and receipt are independently verifiable), is clearly the result of joint creation.

Such developments will make a confrontation with a theory of provenance which forbids joint creation inevitable. In fact, we have encountered such phenomena already, but our theory blinded us to their implications. The monstrous lengths to which our theory of organisations drives those who uphold it is given by Scott -

An example would be the Commonwealth Electoral Office, Tasmania (CA 969), which also undertook functions on behalf of five other agencies. The agency title is given thus:

(1) Commonwealth Electoral Office, Tasmania 1903-1948
[Also] (2) Public Works Branch, Tasmania 1903-1932
(3) Works Registrar, Works and services Branch/ (by 1947) Department of Works and Housing 1932-1948
(4) Deputy Commissioner of Old-Age Pensions, Invalid Pensions, Maternity Allowances and War Pensions/(by 1920) Deputy Commissioner of Old-Age Pensions, Invalid Pensions and Maternity Allowances, Tasmania by 1918-by1943
(5) Deputy Public service Inspector, Tasmania 1903-by 1936
(6) Public service Inspector, Tasmania by 1936-by 1948. 35

Whatever this beast may be, it clearly is not the `Commonwealth Electoral Office, Tasmania, also ...'. One does not have to look far to find the model for this kind of administrative unit -

POOH. ... When all the great officers of State resigned in a body, because they were too proud to serve under an ex-tailor, did I not unhesitatingly accept all their posts at once?
PISH. And the salaries attached to them? You did.
POOH. It is consequently my degrading duty to serve this upstart as First Lord of the Treasury, Lord Chief Justice, Commander-in-Chief, Lord High Admiral, Master of the Buckhounds, Groom of the Backstairs, Archbishop of Titipu, and Lord Mayor, both acting and elect, all rolled into one. And at a salary! A Pooh-Bah paid for his services! I a salaried minion! But I do it! It revolts me, but I do it!36

Although the administrative unit concerned has a single personality, this in no way prevents the exercise of each functional responsibility independently -

NANK. But whom did you get that from?
KO. Oh, from Pooh-Bah. He's my Solicitor.
YUM. But he may be mistaken!
KO. So I thought; so I consulted the Attorney-General, the Lord Chief Justice, the Master of the Rolls, the Judge Ordinary, and the Lord Chancellor. They're all of the same opinion. Never knew such unanimity on a point of law in my life!37

What we have here is a single administrative entity (Commonwealth Pooh-Bah, Tasmania, or something of the kind) exercising several different functions simultaneously. Whether we describe such beasts as one administrative unit or several does not matter very much - it would be nice to have a theory which supplies an answer which we could apply consistently or by which we could, at least, explain departures from normal practice, but it is not critical. The main thing is that we retain the ability somehow to attribute, correctly and unambiguously, records made to one or more identifiable administrative units or agents of some kind which are themselves adequately described and documented.

The convolutions in Scott's 1980 Tasmanian example are clearly being driven by some other more powerful need - viz. the need to delineate separate functional responsibilities as part of the provenance statement . Whatever else this powerful drive is saying, it tells us, in clear and unmistakable terms, that it will not be adequate to our purpose to try to say all we want to say about the context of records by reference simply to an administrative entity (regardless of how it is defined) which we identify as its creator. The muffled undertone you hear as you read over the tortured phrasing of that Tasmanian aberration is the moaning of trapped functions trying to get out. Ideas about function are too important to be subordinated as descriptors of records-creating agents. Functions are not aspects of the life of a records-creating agency. Agencies are episodes in the life of a function. Just as we learnt to dissociate data on provenance from data on recordkeeping, so we must learn to dissociate data on function from data on agencies/persons to express a variety of ideas about provenance.

An agency's functions include not merely its mandate or assigned responsibilities, but also its recordkeeping activities. The fact that an agency could not just be a records-creating entity was apparent to Scott -

In the best of all possible worlds, the agency which produces the records (`creating' or `recording' agency) would also be the agency which transfers them to archival custody ... However, with administrative changes becoming more frequent and complex, records created by one agency are increasingly transferred ... by a successor agency or a number of successor agencies ...

... This distinction between `creating' and `transferring' agency was introduced ... in 1953 by Mr I. Maclean ...

In a further review of our accessioning procedures ... in 1965-1966, it was apparent that to the existing distinction between `creating' and `transferring' agencies we needed to add a third concept : that of `controlling' agency ... where a `transferring' agency had itself either been abolished or its functions transferred to a successor ... now responsible for accessions in archival custody ...

While all three concepts, `creating', `transferring' and `controlling', have valid connotations and application to archival work, it remains our view that, for basic arrangement, description and reference, the `creating' agency is pre-eminent. In this we reflect the views of Jenkinson who recommends classing archives `under the Administration which actually created them' ... 39

Now, whatever one thinks of the view that `for basic arrangement, description and reference, the `creating' agency is pre-eminent', it is clear that by 1980 at least Australian descriptive theory had not yet reached the heart of the problem. Indeed, whether you prefer to formulate a virtual I upon the basis of creation, transfer, control, or any other recordkeeping activity is scarcely an issue worth bothering about. In fact, Australian Archives practice provided for a `records controlled' fonds gathered together in `Agency Manuals' (showing all series currently controlled by an agency regardless of who created them). The fonds based on ideas about records creation were displayed in `Agency Guides' (showing all series created by an agency regardless of when). Similarly, a records-transferred fonds could be compiled from the Accession Register, if that's what you wanted.

The real conceptual difficulty here is that `agency' (i.e. a records-creating body) is being used when documenting recordkeeping activities of transfer and control in addition to creation. There is simply no guarantee that the body which transfers or controls records which it did not create will itself be a creator of records. It is conceptually possible that a body which does not itself fit the definition of agency (independent record-keeping system) will be the one which transfers or controls the records. To follow the logic of Scott's analysis, if the process of records-creation is documented by linking records to records-creating bodies, then the process of records transfer should be documented by linking records to records-transferring bodies, and the process of control documented by linking records to records-controlling bodies.

Of course you won't do that. What you will do, if you're sensible, is redefine agency so that it is conceptually free of any single one of those recordkeeping ideas - so that it can usefully be employed when documenting any of them. The point is that records-creation is only one of the recordkeeping activities which agencies carry out. It is an important one (for some purposes, arguably, the `pre-eminent' one) but you need not (indeed should not) define the entity in terms of the activity which it is the purpose of your documentation to show. Agencies, however defined, must be related to recordkeeping entities in different ways to document each recordkeeping activity. It is the relationship (not the definition) that documents the recordkeeping activity.

The traditional provenance statement (who created the records?) is only a simple, convenient way of packaging up a number of different ideas about records context - just as we used to package up ideas about recordkeeping and context, also for convenience. When it became clear that ideas about recordkeeping and context are better treated separately (because they lead separate lives of their own apart from the period of shared association) we developed methodologies for documenting them separately and re-assembling them according to the relationships our system establishes between them for the period of their association. Ambience and provenance are not, in other words, characteristics of entities but of relationships between entities. The provenance of records is established by showing relationships to (rather than an identity with) their context. Provenance defines certain kinds of relationships between records and contextual entities. Ambience defines relationships between contextual entities. Thus, a function shows provenance when related to records and ambience when related to agencies. An agency shows ambience when related to a subordinate agency and provenance when related to records.

We need to look at unpackaging important ideas about provenance into statements about separately documented contextual entities and the relationships established between them and with records. Just as we have to unravel the personality of Arthur Wellesley from the identity of the office of Prime Minister which he held as successor to and predecessor of other personalities who held the same office (because he, they and it all had an existence apart from each other), so we must unravel the `personality' of corporations from the identity of the corporate package in which they existed for exactly the same reason (because the corporation and some at least of the features of provenance which they embody have an existence separate from each other).

In the same way that it is limiting to confine Arthur Wellesley's role in relation to records by documenting a relationship only when he can credibly be identified as sole creator, you cannot adequately document recordkeeping activities by establishing relationships only with records-creating agencies. If you want to depict `control' you must find an agency to give it to. But, if the administrative unit which exercises control has no independent recordkeeping system, you must either accept that your concept of agency is inadequate and register as an agency a unit which does not create records, or you must misrepresent the situation by attributing control to an agency which is different from the administrative unit which actually exercises it.

Archivists need to identify many significant relationships between records and agencies. In addition to the three identified by Scott (creation, transfer, and control), we may need to say which agency(ies) possessed the records, or owned them, or maintained them, used them, or disposed of them. There must be many others and some of these impinge on the notion of provenance. They certainly say important things about the records-making process.

The information needs of archivists and their users cannot be met by answering these and other recordkeeping questions solely in terms of `which records-creating agency created these records?' The relationship itself must be conceptually separate from the information documenting the two things between which a relationship is shown. Until a relationship is established, it is not possible to speculate about how an agency operated in respect of a set of records. An agency is not a records-creator until we say it is, and we do that by establishing a relationship - not by defining it thus.

The strategic implications of this should be clear, if only because they have been suggested often enough by David Bearman40. While archivists remain trapped within their theory of provenance, they are condemned to invest significant amounts of time in researching and documenting administrative histories which are useful to them in only very limited ways and to others scarcely at all. Freed of our limited conception of what a corporation or person is, our interests in documenting their activities align more easily with work being done by others.

Ways of reducing the burden of researching and keeping up to date vast quantities of contextual data - so often used to criticise the Australian descriptive system - by allying ourselves and joining our activities to those of others (or using the results of their labours), then suggest themselves. Data on government agencies is to be found in numerous compilations (which are usually sources for archivists who reorganise it into finding aids); these include government directories, government information services, and telephone directories. Similar data is used in a variety of government functions : program budgeting, public sector management, managing machinery of government changes, managing freedom of information, and in government information locator systems. In the private sector an important body of data on corporations exists in the national company registration scheme and in business directories. Data on persons is available from the national biographical dictionary and the vast compilation of data by genealogists. We might make more use of such data sources or enter into partnerships to share our data with them.

I do not mean to suggest that this kind of data can be used indiscriminately for descriptive purposes and without regard to quality control. We will get nowhere using lousy data - and much of the data around which we might consider using in archives work is lousy. Problems of data quality represent real obstacles to data sharing. The point is that by concentrating on relationships as the primary tool for establishing provenance (and other archival ideas) we are freer to use someone else's conception of an agency even if it is not the same as ours. Time now spent crafting the boundaries of an agency (or fonds) to the contours of an `organic', specific creation can be spent instead on establishing those relationships which express the ideas we wish to document and, since the primary purpose of archival information systems is to provide external validation for archival data used in documenting recordkeeping systems, on quality control.

Thus provenance may be shown by attributing joint creation of equal value or by assigning primary creation to one agency while another is shown as `associated creator' to indicate a subordinate provenance relationship. It may be expressed in terms of relationships with agency(ies) and the person who `produced' archives of an agency or the functions of the agency(ies) in furtherance of which they were produced. The provenance statement may be expressed to include recordkeeping ideas other than `creation' (e.g. control, transfer, or use).

Records-creation (if you think about it) is really a very imprecise and rather boring idea. It is useful primarily for the purpose of identifying an historical fonds. Throughout this essay, I have spoken of provenance as an historical idea - we identify the provenance of records which have already been created - because that is how archivists traditionally think of it. Electronic recordkeepers are now telling us that they will require provenance data to be available as part of the records-creation process41. It is manifest that such archival data will need to be formulated necessarily in ignorance of whether or not the identified agent exists, definitionally, as the single records-creator of an object of archival description and certainly in ignorance of what records will be created.

Because archivists have always needed to use provenance to express ideas other than creation, our provenance statements have always had to bear the weight of additional, more complex, more useful ideas. It is time those ideas were unravelled and given proper recognition. When one considers the many recordkeeping activities which might usefully be documented by establishing relationships between agents and records, it seems inappropriate to keep on giving records-creation the prominence it has had in our definition. It may well be that the defined entities we use for recordkeeping purposes will not look very different from those we use now. It is clear, however, that we should be prepared to use them more extensively in undertaking a wider variety of documentation tasks and to make use of data on corporations and persons conceived outside of our pre-occupations about records-creation. It will assist us to make this transition if we can stop thinking of corporations and persons conceptually in terms of only one of the uses to which they can be put.

Figure Three

The next stage in our evolution will be to recognise that other kinds of entities can be used to demonstrate provenance - e.g. `organisations', families, and functions. T R Schellenberg (I believe) anticipated this approach forty years ago - Figure Three42. His focus was on recordkeeping, we might now want to fiddle with his diagram (by interchanging `classification scheme' and `organizational unit', for example), and his terminology is not what we would use today, but in essence he argues for a subdivision of ideas about records which conceptually separates function from corporate entity - I hope to say more about this in another place. The essential point to note is that Schellenberg divides `structure' (of records) from `substance' (of context) and within substance he identifies the separate ideas of `organizational units' (corporations), `functions ...' (functions) and `subjects'.

Problems with provenance cannot be solved by using conceptual tools which enable us only to relate a multitude of contextual entities to a multitude of recordkeeping entities. We must have tools which enable us to relate a multitude of contextual entities to a multitude of recordkeeping entities in a multitude of ways. Having good provenance will depend utterly on how well we define and how well we use the (necessarily selective) array of entities and relationships which we employ to accomplish the task. That these tools may also be used to provide `access points' is undoubted - indeed, the need for provenance is another kind of `access' need - and most (if not all) of the contextual and recordkeeping entities we end up employing may be developed primarily for information retrieval purposes. But the information need for good provenance data must never be confused with the need for subject retrieval. From an archival point of view, the primary purpose of a provenance statement is to provide an externally verifiable context for documented recordkeeping activity. A provenance relationship must not be mistaken for a subject relationship. Recordkeepers may use the same tools as information managers but they must use them differently. Archivists, as they explore the ramifications of their ideas about provenance, will forget that at their peril.

ENDNOTES

1. In this article, I shall use the words "corporate' and "corporation" broadly to refer to any group acting as one. this should be distinguished from the legal person-hood of "corporate bodies".
2. International Council on Archives, ISAD(G) : General International Standard Archival Description (Ottawa, 1994), pp. 7-8.
3. Terry Cook, The concept of the archival fonds: theory, description, and provenance in the post-custodial era in The Archival Fonds: From Theory to Practice edited by Terry Eastwood (Bureau of Canadian Archivists, 1992), p.40.
4. By Australian practice, I mean the CRS or `series' system. I do not mean to imply that it is universally applied in Australia.
5. Margaret Hedstrom, Descriptive practices for electronic records : deciding what is essential and imagining what is possible Archivaria 36 (Autumn 1993), pp. 56-57.
6. See endnote 9.
7. Terry Cook, The concept of the archival fonds ..., p.46.
8 There is, of course, literary warrant for such an idea in the very memorable Chapter XXXVIII (in which England is Ruled By An Orange) of Sellar and Yeatman's 1066 and All That (1930) -

Williamanmary for some reason was known as The Orange in their own country of Holland, and were popular as King of England because the people naturally believed it was descended from Nell Glyn. It was on the whole a good King and one of their first Acts was the Toleration Act, which said they would tolerate anything, though afterwards it went back on this and decided that they could not tolerate the Scots

and, perhaps most presciently of all, in Gilbert and Sullivan's own The Gondolier's (1889) Act II -

GUI	. ... It is arranged that, until it is decided which of us two is the actual King, we are to act as one person.
GIORGIO.	Exactly,.
GIU.	Now, although we act as one person, we are, in point of fact, two persons.
ANNIBALE.	Ah, I don't think we can go into that. It is a legal fiction, and legal fictions are solemn things. Situated as we are, we can't recognise two independent responsibilities.
GUI.	No; but you can recognize two independent appetites. It's all very well to say we act as one person, but when you supply us with only one ration between us, I should describe it as a legal fiction carried a little too far.

For "rations" read fonds. See also endnote 21.
9. Peter Scott introduced the notion of simultaneous multiple provenance into his 1966 American Archivist article (see endnote 10) at p.502 when describing how persons fitted his model by providing `the necessary context for personal or family archives and for archives of an agency that may have been produced by a given person' [my emphasis]. He clearly anticipates attribution of a series simultaneously to both a corporate and a personal provenance, though not to more than one corporate agency. This suggestion seems to have gone nowhere, and it is at about this time, according to Scott's later account, that Australian Archives turned its back on this road and took a more traditional view (see endnote 14). For as long as I knew him, Peter continued to worry away at the problem of integrating personal and corporate provenance. So far as I am aware, he never solved it fully to his own satisfaction. The method chosen, establishing a direct person-agency link and thereby an indirect link between series created by that agency and the person, is similar to the indirect provenance established with superior agencies and organisations (see endnote 22).
10. Peter Scott, The record group concept, American Archivist Vol.29, No.4 (October, 1966), pp.493-504.
11. Peter Scott, The record group concept, p.501.
12. Peter Scott, The record group concept, p.502.
13. Now Australian Archives.
14. P J Scott, C D Smith and G Finlay, Archives and administrative change - some methods and approaches (part 3) Archives and Manuscripts Vol.8, No.1 (June, 1980), pp.46-50.
15. James Thurber, Here lies Miss Groby in The Thurber Carnival (New York, Modern Library, ?1957), p.52.
16. Hilary Jenkinson, A Manual of Archive Administration (London, 1966), p.101.
17. This will seem unduly harsh to many of my colleagues, at home and abroad. Let me water it down to this extent. Archivists probably have a fairly sophisticated set of ideas about provenance. What I am on about here is questioning how well they employ those ideas descriptively. It is no good having complex notions about provenance, thinking and writing about it in sophisticated terms, even conceptualising about it theoretically. That complexity must be translated into descriptive tools which are crafted to document complex ideas about records-creation. I think I will stand by my conclusion that archivists have not done this well.
18. Heather Macneil, The context is all: describing a fonds and its parts in accordance with the Rules for Archival Description in The Archival Fonds: From Theory to Practice edited by Terry Eastwood (Bureau of Canadian Archivists, 1992), pp.202-203.
19. Peter Scott, The record group concept, p.498.
20. Chris Hurley, The Australian ("series") system : an exposition in The Records Continuum : Ian Maclean and Australian Archives First Fifty Years Edited by Sue McKemmish and Michael Piggott (Clayton, Vic, 1994).
21. This is not to deny that persons are divisible functionally - just like agencies. This point is made very clearly by Terry Cook in his article "The concept of the archival fonds ..." (at p.42) already cited at endnote 3. The point is that an individual is indivisible personally. See also endnote 8.
22. Both organisations and superior agencies are shown as records-creators indirectly, in the same way as persons (see endnote 9), through the link established with the records-creator. In this way the records creation of a subordinate agency can be seen as the sub-sub-fonds belonging to its superior agency which is itself the sub-fonds of the organisation.
23. David Bearman and Richard Lytle, The power of the principle of provenance, Archivaria 21 (Winter 1985-86), p.16.
24. The government exercised by a feudal prince remained as simple in form as it was limited in object ... As the prince's governmental tasks grew more detailed ... there was a tendency for his court to operate with a more restricted membership of officials and a few barons and bishops, by whom most of the work of administration was performed ... administration was in the hands of a few great officers - the seneschal, chamberlain, constable, butler - their names betray the origins of their offices in household services ... Their duties, not particularly specialized, consisted primarily in the supervision of justice and administration, military command, and household finance and organization. At the local level there were also officials who helped to exploit the lord's domain, supervise local administration, and collect revenues. their names varied : provost, bailli, seneschal, chatelain ...
Norman Zacour, An Introduction to Medieval Institutions (New York, 1969), pp.110-111.
25. Jean Durabin, France in the Making, 843-1180 (Oxford University Press, 1991), pp.277-283.
26. Throughout the Middle Ages ... a nation ... was defined by its birth - as, indeed the etymology of the word demands; it is a group of men of common origin, bound together by ties of blood. In the Middle Ages it simply meant race.
How can one in fact determine whether or not men share a common origin? What do the people of the same nation in fact have in common? The legate Tadwin, said John of Salisbury, was of the Teutonic nation ... Throughout the Middle Ages a nation was characterized by language, physique and customs. But analysis of manners and appearance leads to misunderstandings and endless debate. Ultimately, the only definitive `national' characteristic that applies to everyone, learned and ignorant, is language. A nation in the Middle Ages was primarily a language ...
Now we must look at the way in which the intellectuals debated the nature of the ideal community. The Augustinians thought that there was no relationship between the political and racial communities, populus and natio. But there were few Augustinians. And all those who were tainted to a greater or lesser extent by Aristotelianism did link the two ... Thus most theorists established a close connection between State and race, that is State and `nation' in the medieval sense, which meant above all between State and language.
Towards the end of the thirteenth century rulers began to emphasize the fact that their subjects constituted a `nation' and to make language one of the props of their States ...
Certainly the concept of nation retained an inexact significance for a long while afterwards. It could refer to different entities. Froissart spoke of the nation of London ... Chastellian of the nations of the kingdom of Charles VII ...
The Middle Ages took up and tirelessly repeated all the themes woven by antiquity around the idea of patria, one's native land ... The Digest, heir to a long tradition, recognized two senses of the word patria. Each of the citizens of the Roman Empire had a patria propria (`native land') and a communis patria, which was Rome. With the disappearance of the Empire and the triumph of Catholicism, patria was engulfed in vagueness ...
But then the new States became more forceful. Men became ever more aware of them, accustomed to considering as their country these political entities which increasingly formed the real parameters of their existence. As early as the twelfth century Geoffrey of Monmouth explained that by patria he meant the monarchy of the whole island, totius insulae monarchia. In the second half of the thirteenth century the inhabitants of Liège consciously formed a political unit when they developed common institutions: at the same time they began to speak of the country of Liège ...
Since at the end of the Middle Ages men now considered their State to be their country, all the affective and emotional power attached to the name and concept of the country henceforth upheld the State ...
Bernard Guenée, States and Rulers in Later Medieval Europe tr. by Juliet Vale
(Oxford, Basil Blackwell, 1985), pp. 52-55.
27. ... A professional bureaucracy still functioned in the Byzantine Empire, where anyway the common sense of Roman identity and Christian orthodoxy kept the citizens united throughout a period of grave military crisis. In the west, though, whatever had survived of the apparatus of imperial government into the time of the Ostrogothic kingdom was disrupted by the division of Italy following the Lombard invasion. In Gaul the breakdown of central authority in the fifth century had meant that the Frankish kings inherited little of the governmental apparatus of the Roman Empire ...
What the earlier Merovingians had, but their Carolingian successors seem entirely to have lacked, were instruments of local urban self-government. The survival of such Roman institutions as the municipal archives appears reasonably securely documented into the seventh century, and a number of major urban centres were clearly still flourishing in Aquitaine, Burgundy and Provence at the beginning of the eighth. These, however, were in many cases smashed by the campaigns of Charles Martel and Pippin III ...
Under Charlemagne, as under his immediate predecessors, the governmental apparatus of the state was minimal. Counts still served as the principal royal local officials, but to assist them they had no more than a stipulated minimum of one notary to write the documents they needed. Deputies or viscounts they might have, at least from the early ninth century onwards, but in practice they depended upon goodwill of local landowners for any effective imposition of justice ...
Whatever else his achievements, Charles made no significant improvements to the governmental apparatus of the greatly expanded kingdom he himself had created. The actual central administration remained minimal, and consisted of no more than the ruler's immediate entourage. Those documents that were not left to their beneficiaries to write were written for the king by the clerics of the royal chapel, which thus doubled as the governmental writing office ... Decisions taken at the annual assemblies ... and sets of instructions were not necessarily written or were recorded only by a series of headings rather than in full ...
Roger Collins, Early Medieval Europe 300-1000 (London, Macmillan, 1991), pp.278-279
28. We are accustomed to speak of feudal states, and to the learned in medieval times the idea of the State was certainly not unfamiliar; the texts sometimes employ the old word respublica ... But the idea thus evoked was very different from what it would be today; in particular, it was much less comprehensive.
A long list could be made of the activities which we consider inseparable from the idea of the State, but which the feudal states completely ignored. Education belonged to the Church, and the same was true of poor relief, which was identified with charity. Public works were left to the initiative of the users or of petty local authorities - a most palpable breach with Roman tradition and even with that of Charlemagne ...
For one of the common features of all governments was, not exactly their weakness, but the fact that they were never more than intermittently effective; and this blemish was never more strikingly manifest than where ambitions were greatest and the professed sphere of action widest ...
The best minds clearly appreciated the permanence of the State. Conrad II of Germany is credited by his chaplain with having said : `When the king dies, the kingdom remains, like a ship whose captain has perished.' But the people of Pavia, to whom this admonition was addressed, were undoubtedly much nearer the common opinion when they said that the destruction of the royal palace could not be imputed to them as a crime, because it had taken place during the interregnum ... Prudent persons always persuaded the new monarch to confirm the privileges granted to them by their predecessor ... In other words, no clear distinction was made between the concrete image of the chief and the abstract idea of power.
Marc Bloch, Feudal Society, Vol. 2 Social Classes and Political Organization
tr. by L A Manyon (London, Routledge & Kegan Paul, 1962), pp.408-410.
29. The geographical difficulty of communications implied as a direct consequence that the amount of governing done at headquarters could not be very great: therefore a very simple and a very personal organization was all that was required. Nevertheless, it must not be underestimated, for it shows the embryo of a civil service, the origin of many offices of state and public departments, and it represents, however crudely, an administrative system which could cut right across provincial and local boundaries.
The king's household was essentially domestic both in its nature and in the way it worked. Private servants like the chamberlains and the steward and the butler were responsible for its running. They grew in importance with it, and by Athelstan's reign, before which we know little or nothing about them, they had become sufficiently dignified to witness the royal charters, and by 1066 they were well on their way to being transformed into public officials. From the point of view of future developments the household administration is most usefully considered according to its functions rather than its officials.
The Chamber ... was the king's private apartment, his bedroom, placed in the care of the chamberlains. Off it was the wardrobe where were placed the royal robes as well as the treasure chest which contained the bullion and important records ... The Chamber, the Wardrobe, and the Treasury were all destined to expand into high departments of state. Sooner or later finance came to underlie all their activities ...
[The Chancery] This writing office developed from the organization of the royal chapel, for it seems clear that the royal chaplains, responsible for the daily mass, were doing the king's secretarial work ..,. The Anglo-Saxon chancery exhibited an inventiveness to which there was no parallel in western europe except in Merovingian and Carolingian Gaul and, apart from the many official copies of new laws for transmission to the shire courts, it issued a truly remarkable number of documents ... Evidently, therefore, the work of chancery was being done by a group of highly trained clerks and, though it is not until Edward the Confessor's reign that we find the chief clerk apparently styled `chancellor' and have our earliest surviving specimen of the royal seal, the first `great seal' known in the West, it is evident that the Anglo-Saxon chancery was by then `an ancient and sophisticated institution', the product of a hundred and fifty years' organized development.
G O Sayles, The Medieval Foundations of England (Methuen, 1950), pp. 172-174
30. The aspect of new legal ideas which is often supposed to have had the most impact on lay politics in the twelfth and thirteenth centuries is the derivation from Roman law of the idea of groups with the legal capacity of individuals - what modern Roman law calls legal persons and English common law calls corporations ... Collective activity and governmental repression were matters of politics, not of legal conceptualization : long before academic lawyers had started to get interested in what Roman law texts said about collegia and universitates collective groups had been acting without legal impediment ... Moreover the Roman law texts did not actually define collegia, for the Romans themselves, according to one recent historian of the subject, did not have any clear concept of what their modern successors call legal personality. The chief reason for the persistence of the belief that corporation theories developed in the thirteenth century seems to be the assumption that there is now a single, coherent, and self-evident concept of a corporate body, and that lawyers then were trying to find it. When Innocent IV, for instance, is found making a remark which sounds more or less compatible with modern ideas, he is applauded as if he were `getting warm' in a game of Hunt the Concept. But the idea of Hunt the Thimble only makes sense if a single thimble is there all the time, if one is trying to find it, and if there is someone who has hidden it and knows where it is ... The concept of a legal corporation or legal personality can exist only within a legal system in which there are things which an individual or a corporate group can do and suffer at law which an unincorporated group cannot. It also needs circumstances in which people feel a fairly serious need to distinguish the responsibilities of individuals from those of the groups to which they belong.
Making these distinctions is not so much a matter of intellectual subtlety as of having the occasion to apply subtlety in a particular way. There seems little reason to believe that early medieval people could not distinguish, in a common-sense sort of way, between the rights of a church (or its saint) and the rights of its clergy as individuals, or between the rights of a town and its burgesses. From the twelfth century philosophers and jurists began to skirmish around the difficult edges and implications of the common-sense distinction ... This vagueness was not the result of a failure to make a distinction which was always there but which people at the time failed to see or to express clearly. People did not `confuse' the rights and responsibilities of groups and their members : for most legal purposes the two were the same ...
Susan Reynolds, Kingdoms and Communities in Western Europe, 900-1300
(Oxford, Clarendon, 1986), pp.59-61.
31. The economic transformation of the High Middle Ages was accompanied by far-reaching changes in political and social organization, and in mental attitudes. Europe evolved during these generations from a pre-literate to a literate society. While it is true that most Europeans of 1300 could not read (or at least not very well), they had nevertheless come to depend on written records - deeds, letters, government surveys - to define their rights, property, and status ... The production and preservation of government documents increased spectacularly : surviving papal letters number about 35 per year around 1100 but rise to 3600 per year by the early fourteenth century, and the same explosion of paperwork occurred at royal courts. ...
Taken together, these new records bear witness to increasingly effective and complex royal administrative systems which by 1300 were evolving into modern states. All across Europe, skills such as reading, writing, and mathematical calculation were becoming vital to the functioning of secular and ecclesiastical governments, urban businesses, and even agricultural enterprises. Possessors of these skills, the reasoners and reckoners, sifted into positions of control throughout society, changing its attitudes and its character ...
C Warren Hollister, Medieval Europe, A Short History (New York, John Wiley, 1964), p.149.
32. In the course of the second feudal age political authority, which up to that time was much subdivided, began everywhere to be concentrated in larger organisms ... and a list of [the] causes could almost be compiled by taking the opposites of those which earlier had led to disintegration.
The cessation of the invasions had relieved the royal and princely powers of a task which exhausted their strength. At the same time it made possible the enormous growth of population ... which ... not only facilitated the maintenance of order, but also favoured the revival of towns, of the artisan class, and of trade. As a result of a more active and abundant circulation of money taxation reappeared, and with it salaried officials ... Thus the State from this time onward began to acquire that essential element of its supremacy - financial resources incomparably greater than those of any private person or community. Corresponding changes took place in the mentality of men. The cultural `renaissance', from the end of the eleventh century, had made it easier for them to understand the social bond - which is implicit in the subordination of the individual to the government.
Marc Bloch, Feudal Society, Vol. 2 Social Classes and Political Organization
tr. by L A Manyon (London, Routledge & Kegan Paul, 1962), pp.421-422.
33. To match the concentration of power at provincial and local level, central authority was becoming more unified and streamlined. In the later seventeenth century the most diffuse central authority was that of Spain, where every branch of the state was presided over by a council responsible only to the king. There were councils for each province ... and for each department of state - war, finances, the Inquisition. This machinery reflected the growth in the state's activities over the years : the standard solution to an administrative problem was to set up a council. Similarly in the Austrian domains ... In such states, the complexity of central government led to cumbersome, slow, and inefficient administration. These were drawbacks which, it was increasingly argued, modern states could not afford.
As in provincial administration, France offered an impressive alternative model. The French administration was run by individual ministers : the chancellor or keeper of the seals (justice), the comptroller-general (finances), and the secretaries of state for war, the navy, foreign affairs, and the king's household ... Policies were made by councils ... The secretaries of state, however, had the executive authority in implementing policies ... the general disillusion which characterized Louis XIV's last years had produced much criticism ... When Louis died, a new and more Spanish-looking system, called Polysynodie, was introduced ... Their inexperience told; the councils became scenes of undignified squabbling, and by 1718 the experiment had been abandoned. The power of the secretaries re-emerged, and was never threatened again until 1789. The complaints of `ministerial despotism' so common during the crisis which culminated in that year, showed how dangerously effective many Frenchmen believed this power to be.
William Doyle, The Old European Order, 1660-1800 (Oxford, 1978), pp. 257-259.
34. Terry Cook, The concept of the archival fonds ..., pp.61-62.
35. P J Scott et