I am inspired to write this blog after reading an exceptional blog from Michaelle Haughian entitled “2b or not 2b, that is the question“, in which she presents a variety of cases that present challenges concerning whether to create a new identity for an item or whether to merely consider the part as integral with the whole and not worthy of unique identification.

I have been considering this problem for many years and I have arrived at the conclusion that “whole-part” structural relationships are a universal attribute of all heritage documentation and that any sufficient documentation system cannot avoid a complete treatment of this problem. This complete treatment requires both standards for identification and numbering, as well as a policy on when to create unique identity for the parts and when it is better to simply integrate the parts within the whole without a unique identity.

As Michaelle explains, there are many situations in which this problem exists, such as the difference between a painting and its frame, a set of coins, and a chest of drawers. From my own experience this whole-part problem is universal with there being relatively few situations in which items are seen as totally distinct and all evidence showing strong structural relationships with other items including items unavailable or unknown (such as when you have one book of a series and wish to note the series and the other books as missing).

Essentially, the whole-part problem concerns when we should identify an item as a whole in its own right, and when should we view this as a part, and to what extent to we treat the part as a whole in its own right, perhaps recursively. I have referred to this as the “tea set” problem, in which there is a tea set with a silver tray, which in turn has a sugar bowl which has a separate lid. The entire tea set is a single unit or whole, and each of the parts are integral to the whole, but can also be identified individually. The sugar bowl, a single part within the tea set, it not only an identifiably part, but is a also a whole which itself contains the bowl and the lid as part, thus having a 3-level structure.

The question arises as to how best to document this for identification and structural purposes and what policy statements should inform this process, and to meet this requirement I have conceptualised three forms of structural relationship between items, which I call the Set, the Unit, and the Part levels..

The Set is not necessarily a “collection” in the common application of this word, but can be any group of items which are seen as a unit and managed as a unit. They are typically acquired, moved and exhibited as a unit, but the items may not have any direct structural relationship with each other. For example a coin set, which consists of a number of coins which are each individually identified, and which allows for the documentation of missing items which may form an element of future acquisition policies. The Set is essentially just another item and is given a single number, and this number can be used for management purposes when referring to the management actions, such as the SPECTRUM procedures, which influence the collection as a whole.

The Unit is an item which is uniquely identified and given a number, and a Part is a part of a Unit for which no such number is given, and it is at this point that the arguments of Michaelle come into being. I have approached this problem by treating the Unit as the atomic level for identification of items in the sense that this is the lowest level at which any of the procedures will apply. In particular I consider the process of conservation or repair of items, and ask whether, for this purpose, the item would be treated as a unit or would be further divided into its parts. For example, for the case of the tea set, damage to the lid of the sugar bowl may require that this lid alone is moved into a conversation or repair treatment, with there being no need to move the rest of the tea set. Perhaps the same reasoning applies to the chest of drawers in which a single drawer may be moved into the furniture conservation area while the remainder of the chest remains in its home location. It is this requirement to separate the whole from the parts which I have used to determine whether to create a new number or not.

Thus I see the Unit level as the level of process, in which unique identity is required to support administrative documentation and control, so that consequently the Part level is used specifically to create a finer grain-size for documentation, such as for describing the types of materials, colors, conditions, etc… of  Part of a Unit as distinct from the Unit itself.

Another facet of this argument is that the decision to create unit numbers as distinct from the whole can be delayed until the time that this is needed. This approach is the norm with archival documentation, in which a box will be numbered and then will be described at a coarse level in terms of its general contents, but without allocating unique numbers to the individual items in the contents until this box is further analysed for research or digitisation purposes, perhaps some years later. The same approach can be applied to field collections and other bulk accessions in which the accession takes place at a high level with some delay before the individual items are identified and numbered. This reduces the effort of the accession process and rather pushes this into a retrospective documentation (a la SPECTRUM) task for later

I see evidence of this whole-part problem within not only the physical collections of museums, but also within every other area of historical documentation including events (the liberation struggle in South Africa, of which one part is the Women’s March in 1956), the oral history (in which one collective story is told by many from their own perspective), archaeological sites (divided up into parts for more detailed analysis, such as the Cradle of Humankind being separated into the various caves and sites), and archival materials (in which a collection of an author’s works are presented in boxes, which are divided into files and photograph albums, which themselves are divided into units).

We will not succeed if creating complete documentation systems if we treat collections as a set of units with no structural arrangements, and rather I argue for the treatment of all units as though they are always in whole-part relationships. This is what I presented in the Data Coding Standard for Heritage in 2007, following one year of analysis of documentation requirements in all of the heritage disciplines, and in the intervening 6 years my view on this has not diminished, but has rather been reinforced through further evidence of this universal attribute of history – that everything we collect, analyse, and describe exists in the context of its whole-part structural relationships.