Some thoughts occasioned by the papers contributed to the INTERNATIONAL CONFERENCE ON THE PRINCIPLES AND FUTURE DEVELOPMENT OF AACR
The issue of time underlies much of the discussion about the relationships between Work, Expression, Manifestation and Item (to use the terms applied in Functional Requirements of Bibliographic Records), and is at the root of many of the difficulties associated with online publications, but it is much neglected. So far, only the paper on serials raises it as a substantive issue; the fact that it permeates many of our decisions on every area of cataloguing is obscured by the acceptance of unspoken presumptions.
Catalogue records give the impression that they describe an entire edition (and AACR encourages us in that belief) but most of the time we really catalogue Items. We have a physical object in front of us (book, CD &c) and from that Item we make inferences about the existence of a Manifestation.
Items exist over time. A book with a long history may over the years acquire an interesting provenance, an interesting binding, a manuscript annotation (such as Fermat's last theorem), or have some pages ripped out. Within the library, it may be rebound, may be lent out, or lost. Some of these are transient states, others (once they've happened) are permanent, or at least of long duration. Some affect the Item materially, some do not. A catalogue record (or a MARC record which incorporates library housekeeping functions) often needs to register these states/changes.
A Manifestation may have an abstract existence over time (once created, it's "there" in some sense). If all the physical examples are destroyed, does the Manifestation still exist? I suggest that what we're really doing in cataloguing Manifestations is trying to characterize, by inference, the nature of the physical medium that existed when the Items were produced: the metal or photolithographic plates, the pattern of magnetised particles on a tape or disc, the master copy of a film &c Essentially we're taking a snapshot of the physical source at the moment of production; or rather, inferring the snapshot from its traces in Items.
The instruction to ignore a re-issue of an edition (1.2D3) is the closest AACR2 comes to acknowledging the time factor for monographs. In rare book cataloguing, on the other hand, even minor changes can trigger the requirement for a new bibliographic record, to reflect a changed physical source: we take the snapshots at more frequent intervals.
The creation of online digital resources brings this issue to the fore. The "Item" will be the pattern of bits delivered to your computer when you access the resource, together with what your computer makes of them. The "moment of production", however, is not some moment in the past, but the moment when you demand the resource: and the (re)source can be changed very easily. We as cataloguers can bemoan the instability of online resources, but the problem is not new: it is just that we have been able to fudge the issue till now, because the effort/expense involved in producing a new set of physical objects (Items) reflecting the change in the object underlying the Manifestation meant that it was not undertaken lightly.
At the other end of the technological spectrum, manuscript cataloguers have had to deal with similar problems for years. A manuscript can combine Work, Expression, Manifestation and Item in one physical object: there is no question of separating the Item from its Manifestation. The level at which to pitch elements of the description of the manuscript can be problematical: how many different hands have laboured over how many different time periods to produce the text/picture in the manuscript as it now exists. Even if all is written in one hand, the problems can be intractable.
If we think back to the source of a Manifestation (i.e. the physical plates &c that produced the Item) then that itself will usually be a manuscript/typescript/computer disk &c: and the plates may be set from the manuscript as it existed at a given moment in time. As the Item is a snapshot of the Manifestation, so the Manifestation is a snapshot of the Expression as it exists at a given time. But all of these levels do exist over time. The idea of the "Work" can be seen as an acknowledgement that we need an overarching concept covering the various Expressions that are created and exist over time.
Because library materials can represent Manifestations which themselves represent other Manifestations (e.g. a facsimile reprint) one can have a complex web of interrelationships between the various levels. For a particularly complicated case look at Christopher Tolkien's publcation of The History of the Lord of the Rings, in which he tries to reproduce in one Manifestation the essence of the entire complex of manuscripts underlying the text of the trilogy: he reproduces parallel passages of different drafts of the same text, and attempts to convey the history of individual manuscripts where J.R.R. Tolkien rewrote over existing text (sometimes erasing the earlier text entirely, sometimes erasing it but leaving legible traces, sometimes simply crossing out).
Serials are a case where the Manifestation is not so much a snapshot as a movie.
Dramatic and musical performances often bring out more clearly the snapshot nature: a performance of Macbeth is an instance of how that work was realised in one place on one day. A run of 1000 performances can be seen as analogous to an updating looseleaf publication: each performance replaces the previous day's, and revisions may be made to the performance from time to time.
The oxymoron of the "live recording" is not contradictory in this analysis: each performance is a Manifestation embodied in a single Item, and one of them is then copied as another Manifestation: it is the equivalent of a facsimile reprint in the text world.The studio recording, on the other hand (or the production of a motion picture) is like the composition process of a Work: the Manifestation crystallises at the point where the master recording is created.
I was surprised in Yee's perceptive analysis of "What is a Work" that her description of three layers of creative activity in the creation of a filmed dramatic/musical work omits performance. She lists:
1. The original composition of text/score
2. Production decisions on lighting, sets, delivery,of lines &c
3. Cinematic decisions on camera angles, framing, cutting &c.
In a live recording, performance comes between 2 and 3. In FRBR terms, I would see 1 (composition) as relating to the Work, 2 (production) relating to the Expression, then performance is the Manifestation; in the context of an unfilmed performance then it is also the unique instantiation of the Item; in a filmed performance then multiple Items can result. Activity 3 (cinematic decisions) can involve almost zero creative input (placing a camera in front of the stage and letting it roll), or can involve a more complex creative process. The analysis of the types of derivative relationships (as in Vellucci's paper) is clearly relevant here. At the other end of the scale, a Hollywood movie does not reach a recognisble Manifestation stage until after all the type 3 activities (cinematic decisions) have taken place. No performance as such takes place until it is created in the cutting room.
In summary:
Works exist over time, and the effects of time can be seen by comparing the different
Expressions
Expressions exist over time, and the effects can be seen by comparing the different Manifestations
which reflect the state at a given moment
Manifestations exist over time, and the effects can be seen by comparing the different Items
Items exist over time.
Cataloguing revolves around objects/entities which can have a physical existence over time. We can record that existence continuously (and often do at the Item-specific level, and for serials), or we can take what are essentially snapshots of the state of an object at a given moment in time (for example the editions of a Work). For some objects (e.g. performances) the actual time-span of their existence is brief. The extent to which this is important depends on how that existence impinges on the real world. If a performance is recorded, we can view the performance on different occasions and act as if the original performance still exists. For books, the "performance" is the metal or lithographic plates (or computer file) used to produce the Items: the plates/files may be destroyed/deleted, but the Items remain.
When entities (at whatever level, from Work down to Item) change, they can do so by augmentation (adding more to what is there) or metamorphosis (where something new replaces what was there).
The issues relating to time become more prominent for serials and for online resources, but they affect every aspect of cataloguing. We have survived by fudging the issue while book Mainfestations were the focus of cataloguing, because their production technology allowed us to ignore these considerations; but this has not endeared us to, for example, our colleagues cataloguing manuscripts or rare books with provenance histories.
Functional Requirements of Bibliographic Records largely ignores the time aspect,but I suggest there is much to be gained from an analysis, at each level (Work, Expression, Manifestation, Item) of:
(a) how objects/entities of library interest exist over time; (b) how or whether they change over time (c) how or whether their existence is reflected in some sort of physical reality, tangible or not (d) whether the physical reality is continuous or intermittent
I believe that this would give us a better insight into what are users are really looking for, and what will satisfy their needs, when they conduct bibliographic searches.
M. Heaney
11 August 1997