back to conference papers page

Conference Papers Online

Session 4: Providing a Focus or Limiting a View ?

Can the sea-scape be viewed squarely when the port-hole is round ? Observations on the practical implementation of archival constructs in electronic systems.

Dr Stephen Ellis

National Archives of Australia

I have been asked to help provoke discussion today by addressing the challenges posed by electronic records, and in particular by comparing, in the context of electronic records, the conceptual model for description of archives offered by the Commonwealth Record Series system, developed and used by the National Archives of Australia, with that prepared for the International Council of Archives in Peter Horsmanís Draft Design for an Archival Description System (v. 2)[1] . As this is a conference based on the archives of science and technology, I have chosen my examples with a scientific bent in mind, and also, as you will have divined from the title of my session, with something of a nautical bent.

What I would like to do in the short time available to me is to provide some stimulus to the later discussion we shall have by:

  • first of all making some theoretical observations about the nature of "archival constructs" and what we might expect of them;
  • then by outlining how the National Archives of Australia is implementing the underlying constructs of the CRS system through its current Systems Integration and Redevelopment Project; and
  • finally by comparing our experience of those implementation challenges with the proposals in the ICA draft design.
Let me begin by addressing directly the question our conference organisers have posed for our session: "Archival constructs providing a focus or limiting a view ?" My answer is a succinct, if not brusque YES and YES !

Archival constructs are very much for archivists what theories, hypotheses and paradigms are for scientists: they provide a commonly understood and expressed framework within which further observation and development can be pursued. Very often such constructs provide a basis for devising solutions to real-world problems, but those solutions themselves usually depend on the availability of technologies derived from other hypothesizing, observation and development. As with scientific discoveries and technological progress, there is therefore with archival constructs a complex interaction of theoretical constructs proposing solutions which influence in turn the further development of the underlying theories or paradigms. All of this interaction however, is limited by the very paradigm within which it occurs. Permit me to illustrate my point by drawing upon some actual archives:

At the turn of the 16th/17th centuries the European astronomer Johann Kepler proposed various hypotheses to explain the observed behaviour of heavenly bodies. Keplerís propositions became significant foundations for the further development of modern astronomy. One of Keplerís propositions was that cosmic rays influenced terrestrial behaviours Ė he was after all astrologer to Count Wallenstein in the later part of his life.[2] Astrology as a set of constructs has fallen from grace in more recent times and is now clearly distinguished from scientific astronomy. In the 17th century, however, Keplerís cosmic radiation theories had an equal standing with his now more acceptable explanations of the movements of the planets.

Letís now turn to the less credulous times of our own century and indeed of our own country. In using my next example I will take it as a given in this building and with this audience that we share a reasonably common understanding of modern scientific views on the structure of the earthís atmosphere and the impact on its layers of variations in solar radiation resulting from sunspots and coronal flares.

In the first decades of this century Australia, as a proud, if not arrogant, fledging nation, viewed itself as destined to play a powerful role in the South Pacific. A powerful role requires some means of implementation, and the technology chosen at the time was naval power, in the form of the Royal Australian Navyís fleet unit, [which consisted of the most modern, most powerfully gunned and fastest ships in the southern hemisphere]. When the Australian fleet was on its maiden voyage in 1913 from the heart of the Empire (or from England, for those of you too young to remember) to its future home ports in the south via the Cape of Good Hope, extensive experiments were conducted on the performance of its radio equipment. Observations were made of the strengths and character of radio signals along the whole course taken by HMAS Australia and HMAS Sydney across the southern Indian Ocean between the Cape and Perth. These observations were diligently reported to both the British Admiralty radio specialists at HMS Vernon and to the Commonwealth Bureau of Meteorology. The captain of the Australia thought it was "probable" that the quality of the radio signals had been influenced by "the high barometer (30.5 & above) and the dry state of the atmosphere".[3] The no less credulous fleet radio officer wanted to know not only the barometer readings but also the wet and dry bulb temperature of the air, the state of the sea, the state of the sky, the direction and force of the wind and the sea temperature. The Admiral and the Navy Board and even the Commonwealth Meteorologist thought this was all good stuff and a very commendable effort.

These observations no doubt were not only technologically achievable but also sensible within the prevailing construct for these early implementers of radio technology. We know now of course that all of the factors mentioned by the diligent RAN observers were of no real relevance to the performance of radio transmissions at all. It was not until after the extensive observations of the stratosphere from the 1920s through the 1940s and the discovery of the ionosphere and the magnetosphere that scientists came to a construct which explained the behaviour of earth-based radio transmissions. And when they did so, it was in terms much more consonant with Keplerís cosmic radiation propositions than with the patently obvious earth-bound observations of sensible naval technicians of the early 20th century.

So what can we learn about theoretical constructs from this sort of example ? Clearly they can be useful and are necessary to frame courses of action and further development, but they can also limit options considered by determining the sorts of questions that are asked. These limitations can be made more severe by the technologies available for their implementation, not just because the technologies themselves have limitations, but also because they can obscure essential characteristics of the construct. For example, for over six centuries the basic technology used for the production of records in the western world barely changed, and this stability gave rise to an understandable conflation in most peopleís minds of the technology with the essential concept. This confusion lies at the base of our present difficulties in grappling with electronic records.

Constructs have another related inherent limitation they address what they address and therefore exclude what they donít address. For example, the rules of provenance in traditional archival diplomatics seek to address the real-world problem of the authenticity and reliability of records. Authentic and reliable written records constitute one of the most significant ways in which power is established and exercised in modern literate societies but the rules of diplomatics or provenance have nothing to say about power relationships (and appropriately so). Nevertheless, power relativities in a society may significantly affect the creation, maintenance and accessibility and use of records and archives. No amount of tinkering with different constructs about provenance will address these relativities.

So good constructs are like a craftsmanís tools they are best used to address the real-world problems they have been devised to address. The test of the benefit to be gained from a given construct is its contribution to providing a workable solution to real-world problems in the present state of technology, rather than its comprehensiveness in terms of all possible views. Thus the Commonwealth Record Series system was a set of concepts developed by the National Archives of Australiaís predecessors from the 1960s to address the real-world problem of documenting the administrative context and provenance of records over times of substantial change in the structure of government administration. In our experience it provides an elegant and workable solution to this problem and the advent of electronic systems has provided technology which has made its implementation even more feasible and maintainable.

Another attribute of good constructs, as of good theory, is that they enable you to discover new possibilities and to take practical measures successfully. The classic example is Mendeleevís periodic table of chemical valencies. Our conference organisers have suggested in the outline for this session that "Recent experience with electronic records and the use of functional models have revealed more clearly the nature of archival constructs and the pressing need to identify and define these constructs." I see it otherwise. A measure of how good a construct traditional archival principles have been in the terms I have defined is that they have undergone a re-discovery and renaissance since the emergence of the problems of electronic records. For too long in the early days of considering the challenges posed by electronic records archivists in Australia (as well as elsewhere) tried to solve the apparent problems by dealing mainly with the artifacts presented by the technology. Genuine progress has only come when we have turned back to the fundamental principles of archival science and have concentrated on what constitutes the record itself, rather than on the medium on which it is manifested at a particular time. As I donít have time now to elaborate the proof of this particular point, I invite you to contrast the character and quality of the relevant professional literature of the 1980s with that of the 1990s.

So enough for the theoretical consideration of archival constructs as a type. What does the experience of the National Archives of Australia in implementing the CRS system tell us about its effectiveness in addressing real-world problems ?

As I mentioned earlier, the CRS system is primarily concerned with establishing control over and enabling the effective description of series of records of successive federal governments through extended periods of administrative change. In our experience it does this quite effectively from the point of view of the archivist seeking to establish control for various operational purposes. It does however present significant challenges to the casual user of the Archives because it demands a good understanding of the descriptive structure itself in order to return a useful result. The consequence has been that most users of the Archives have been daunted by demands of the construct as we have implemented it, both because of its inherent complexity and because of the relatively user-unfriendly presentation.

In its current project to integrate and re-develop its electronic systems through which the CRS system has been implemented, the Archives is seeking particularly to address this challenge of accessibility, rather than that of administrative change. Estimates , and methods of estimation, vary about the proportion of the Archivesí holdings which have been accessed by the public, but even the most optimistic places it well below 1% of total holdings. It is increasingly difficult to justify public expenditure on preserving archives the greater proportion of which no one has ever seen since they were transferred to the Archives. Another concern has been that, in common with all archives in this country and elsewhere, we have become sharply aware of the rapid and continuing physical deterioration of the records we hold which are considered to be of enduring value.

At the same time, the public interest in access to the federal governmentís records, as measured by reference enquiries to the Archives, has been increasing steadily over the past decades, and is expected to continue to increase. To cope with this greater interest within the expectation of government that funding will not increase, the Archives must find more effective ways to make records accessible, both through descriptive practices and through the means of delivering the results of searches. So the real-world problem we are currently addressing is the challenge to increase accessibility by orders of magnitude on past practice.

In doing this we are not abandoning the CRS system but rather extending its implementation fully to the record-item level and the governmental functions which give rise to records. The underlying concepts of the CRS system continue to support all the operational needs of the Archives and our present project does not seek to alter that foundation. We are however, concentrating on aspects of the Archivesí operations which are simply not addressed by the CRS concepts, namely on the interactions between the Archives and its clients, both government agencies and members of the public. The following overhead illustrates this position graphically.


The project focuses therefore on the business processes of the Archives rather than the conceptual frameworks underlying them. But there are of course inter-relationships which must be kept in mind, particularly in relation to the description of archives for discovery and retrieval purposes. For example, the CRS system provides substantial standards (consistent with ISAD(G)) for the description of record items and of their administrative context and provenance through the links to series and agencies. In our new system we shall be harnessing also the descriptive and retrieval power of the appraisal decisions in relation to an item through its association with functions of government as controlled through a standardised thesaurus:

  • every item transferred to the Archives from a government agency, in addition to its normal descriptive metadata, will have been sentenced under a disposal authority;
  • every disposal authority will be associated with one or more government functions;
  • searches of the system will use not only the direct descriptive metadata of the items but also the indirect related functional data through the disposal authority link.
So how does this approach compare with that proposed in the ICA draft Design for an Archival Descriptive System ?
The ICA draft design derives from the standardisation work of ICA which produced ISAD(G).[4] It is an attempt to model the entities and relationships in an electronic system which meets the ISAD(G) descriptive content requirements. It commendably places clear emphasis on the role of business transactions giving rise to records in "operationalising" the functions which an agency or organisation fulfills. The ICA draft design in this respect is very close to the modelling prepared by the University of British Columbia/US Department of Defence team (which used similar systems modeling tools based on IDEF0: Integration Definition for Function Modelling).[5] In this approach, the whole of the range of entities and relationships involved in the bringing of records into being has been modelled and is made a part of the proposed descriptive system, extending from the "record-keeping warrant" of the organisation through to its business processes which operationalise its functions and give rise to the records which are the concern of the system. I have no arguments with the logic used in this modelling, but I do have considerable reservations about whether all of the entities and relationships would be or indeed ought to be included within an archival descriptive system.

The Horsman model, if I may briefly summarise it, is an hierarchical construct, relating the functions of the actors or agencies in society to the business transactions through which they are operationalised and to the records which provide the means for , or are the manifestations of, those operations. Although the terminology is different, there are many similarities with the CRS system, with one notable exception the record-keeping system. The CRS model does not explicitly recognise the record-keeping system as an entity within its model. It is rather indirectly or implicitly recognised through the descriptions of the system of arrangement within series and through the relationships linking controlled and controlling series and controlling agencies. Peter Horsman proposes that the record-keeping system of the originating agency be recognised as an entity within the archival descriptive system.[6]

The explicit recognition within the construct of record-keeping systems and of organisational functions has been advocated by a number of other observers and commentators on the Australian series system such as Chris Hurley. I think the key issue here is not so much whether such entities need to be recognised as whether they can be effectively recognised in the implementation in ways which add value for descriptive and retrieval purposes commensurate with the cost of maintaining such metadata. I think they can, and I think there are a number of ways to achieve that, one of which would be the way outlined in the ICA draft.

There are some aspects of the ICA draft design which I find rather puzzling.

In outlining the functions of the proposed system which he is designing, Peter Horsman makes this rather strange statement:

First of all: a description system is not a retrieval system. The system supports the archivist in his archival task, not directly the researcher in his work.[7]
I find this hard to accept in a practical sense surely no institution would be prepared to put in the work required to get such a system up and to maintain it unless it did provide a multiplicity of real-world benefits, one of which would surely be retrieval ?

At a conceptual level, the model also concerns itself with the concept of the fonds, "(as a logical concept) reporting all the descriptions of items created by one single person or organisation."[8] It is probably some sort of heresy, but I really find it difficult to understand what practical use the concept of the fonds so defined can possibly be. It seems to me to be some sort of definition of a universal in Platonic terms which does not really clarify anything much in particular. Nor do I see how a descriptive system could hold such a comprehensive body of data for the records of an organisation unless it sought to encompass the functions of the record-keeping system without limit of time.

The most serious reservation I have about the model as drafted concerns this blurring of the lines between the archival descriptive system and the record-keeping system of the agency. This is most clearly expressed in the following passage:

A Record may change an Archival Item (result of a record-keeping action). In this case at one moment a Record would only be part of one Archival Item. Depending on the chosen definition of Record and its relationship to the Business Transaction, even one logical record might be recorded at one time on more than one Archival Item. An example: if a complete financial transaction, such as the payment of a purchase would be considered as one Business Transaction, the recording of this will likely be scattered over more than one Archival Item in the book-keeping system (daybook, ledger, etc.).[9]

I donít see this sort of detailed elaboration of the transactional relationships between archival items as being a function of the archival descriptive system at all, but rather one of the agencyís record-keeping system. It would seem to me that different record-keepers might make differing decisions about how they wanted to record or relate business transactions in their record-keeping systems, and it is the task of the archival descriptive system to describe what they did, not to create or to realise relationships which otherwise may have been merely potential. These proposals seem to me to blur the distinction between the functions of the original record-keeper and those of the archivist.

In saying this I am not seeking to controvert the very real advantages to be gained from recognising the continuity between those sets of functions, but I think you can go too far to be of practical use, as some of the more recent writing on the "records continuum" model has, in my humble opinion. Perhaps this is another "archival construct" which we ought to discuss in our session.[10]

For now, let me bring us back to the question of how archival constructs can help with the description and management of electronic records as archives. We appear here to be presented with the contradictory demands of preserving access to both "manipulability" and "recordness". The simple answer is that we canít do both simultaneously and you would seriously have to question why anyone would want to manipulate the essential elements of a record itself, rather than perhaps to manipulate the means by which it was accessed or manifested. We must concentrate on the nature of the record as authentic and reliable evidence of a transaction. In my view the major challenges facing the archives profession now are those associated with ensuring the authenticity and reliability of electronic records over time. To do this we need to go back to the long-established archival constructs about what constitutes authenticity and reliability and to re-interpret them in the light of contemporary technology. This is the point where I hope to draw some tenuous connection with the title of my session:

Port-holes are round not because of the functions they perform, which are to let daylight into ships and to permit people to see out of them. Their shape is determined by metallurgy, the physics of stress raisers in the crystalline structures of steel plating, and the methods of application of steel technology to naval architecture. Wooden ships had square windows round ones are harder to make in timber. But square perforations in steel plates create points at the corners where any stress forces on the plate are multiplied, breaking the crystalline bonds within the metal and propagating stress fractures of the plate. Stress fractures sink ships even more surely than loose lips. So when the new technology of steel sheeting came to be applied to the real-world problem of increasing the size and durability of ships, the design of portholes had to be modified to cope with the characteristics of the new technology. The function of portholes did not change.

Thus we are at the cross-roads of implementing new technologies for the creation and maintenance of records. The function of records to provide authentic and reliable evidence of organisational and personal transactions has not and will not change. We need to find ways compatible with the new technologies to enable those functions to continue to be implemented in the real world.

stephen.GIF 54.99 K

[1] International Council on Archives Committee on Archival Automation, September 1996; currently available on the Conference web site:
[2] This application of Keplerís celestial constructs was not exactly propitious for either astrology or for the Count, who was eventually murdered by members of his own bodyguard.
[3] National Archives of Australia (Melbourne): Navy Office; MP472/1/0, file 16/14/3942; Extracts from deck log, HMAS Australia.
[4] General International Standard Archival Description, ICA, Ottawa, 1994. The standard is currently accessible at
[5] See Luciana Duranti, Terry Eastwood, Heather MacNeil: The Preservation of the Integrity of Electronic Records, currently available at
[6] ICA Draft Design, esp. p. 14.
[7] ibid., p. 5.
[8] ibid., p. 6. In the conference session discussion I pointed out that Jenkinson had raised the problem of the concept of the fonds in the first edition of his Manual. Re-reading his points, I realised that in a conceptual model based on the series, the fonds is entirely superfluous for all practical purposes.
[9] ibid. p. 13. In the Conference session, I spoke briefly to overheads of the entity relationship diagrams in the Horsman model to illustrate these points.
[10] In the Conference session discussion, I expressed my personal opinion that the elaboration of the concept of the continuum to encompass various phases and dimensions of being was a complication of essentially simple archival issues, which does not add anything of practical use to the approach when applied in the context of electronic record-keeping systems, but does add unnecessary, and in my opinion spurious, complexity.

back to conference papers page

Published by: Australian Science Archives Project on ASAPWeb
Comments or questions to: ASAPWeb (
Prepared by: Helen Morgan
Graphics by Lisa Cianci
Date modified: 7 October 1999