The effective classification of material—the recognition and exposition of the order within its diversity through the definition of recognised types and sub-types and the devising of accurate terminology to label them—is a fundamental basis of research into material culture. Accurate and rigorous classification (or 'taxonomy') depends not only on the recognition of similarities and distinctions between items but, crucially, also on an informed understanding of which traits are significant (so it is value-based: see Section 4.3.1). Classification is thus inherently analytical. The multi-disciplinary nature of carved stone research raises special problems for effective taxonomy as different disciplines (e.g. art history or archaeology) may have different ways of referring to the same material. Distinctions which are relevant to one discipline may be less so to another. Terms may have different meanings in another discipline (e.g. 'inscribed') or not be understood at all (e.g. 'orthostat'). Furthermore, terms which have popular currency may be too imprecise for scholarly use, while terms used by researchers may have no resonance with the general public (e.g. 'cross' versus 'cross-slab', 'cross-marked stone', 'free-standing cross', etc.).
Classification is not an arid academic exercise. On the contrary, taxonomy inevitably shapes thinking as it focusses attention on, or diverts it away from, certain types of material, or aspects of a particular type. It thus has a profound effect on which kinds of material are considered together and which are not, indeed which kinds are considered at all. The contrast in the level of attention given in Ireland and in Scotland to what appears to be a shared class of material but which is referred to by different labels—'bullaun' and 'rock-cut basin' respectively—is a case in point (Bullaun stones: Case Study 2). It also raises the question of the extent to which it is possible and desirable to align the classification of Scottish material with the terminology used of the same types of material in other countries.
The detrimental effect of a no-longer useful taxonomy is illustrated by the case of the deceptively clear and simple three-class system for analysing Pictish sculpture set out in ECMS (Allen and Anderson 1903). The clarity and simplicity of the system, and the authority of its creators, has led it to dominate discussion of Pictish sculpture for over a century. Its shortcomings have been increasingly highlighted by expert scholars (Henderson 1993a; though for a partial defence see Clarke 2007), yet general academic and public writings continue to perpetuate it. It elevates the presence or absence of Pictish symbols over other, arguably more relevant, criteria, resulting in a distorted impression of the material: for instance by splitting what is arguably a unitary type—the upright cross-slab—into two sub-groups based on whether or not they bear symbols. Conversely, by lumping together all non-symbol-bearing relief sculpture into the overly heterogeneous Class III, the significance of this diverse body of sculpture was obscured. Allen and Anderson's scheme had no place for simple, cross-incised stones and this numerous and important type was largely omitted from ECMS. It was not until Henderson's pioneering work in the 1980s, and her extension of the Allen and Anderson scheme to a new 'Class IV' label, that these stones were properly recognised as a category and integrated into research (Henderson 1987a).
Classification is not a once-and-for-all activity—as understanding evolves, so does the taxonomy based on it, and as understanding changes existing terminologies may need to be refined, or even abandoned altogether. It may take time before such innovations are accepted within a discipline, whether from inertia or because new interpretations are actively contested. Even when scholarly consensus is reached this does not necessarily ensure speedy adoption by wider audiences, as the dogged longevity of the ECMS system demonstrates. The tension between keeping abreast of the latest terminological developments and sticking to authoritative terms that are widely understood is particularly acute for resources (such as Canmore) which have a wide range of users. Faddish or controversial systems which lack widespread support within the scholarly community must be avoided, yet, as the ECMS example above shows, simply because a system is long hallowed, does not guarantee its usefulness any more than does its novelty. One of the advantages of online resources over physical publications is that, in theory, they can be readily updated to take account of evolving classifications and terminologies, and items can be cross-referenced with multiple tags to ensure that they can be found using a range of alternative terms (and these can be embedded to avoid them cluttering up an entry). In practice, however, resources for this level of indexing are currently lacking.
Existing classifications and terminologies have evolved organically within period specialisms and generally represent the cumulative efforts of a number of scholars, sometimes over generations (N.B. study of past classification systems can be a useful means of exploring past value). Only exceptionally are they the result of initiatives by an individual or committee to set out an over-arching schema (typically in association with the preparation of a corpus or major catalogue). Earlier taxonomies tended to prioritize the content of the carving rather than a more archaeological approach prioritizing form and function (e.g. the classification of Iona slabs on the basis of form of cross-motif carved in them (RCAHMS 1982) versus a classification based on monument form (as a proxy of function) and geology (as a proxy for date) (Forsyth and Maldonado in prep). The complex, multi-faceted nature of carved stones, and the diversity of approaches to their study means that there is unlikely to be a single, perfect taxonomy which satisfies all users. Different systems will have their merits depending on the purpose to which they are put. The key is rather to be explicit about the principles of classification, to use labels consistently, and not to diverge from existing systems without clearly articulated justification. Similarly, cumbersomely long-winded labels which are accurate but inelegant, or which strain normal English usage, are unlikely to gain support.
The long-standing challenges of classification and labelling have taken on a new dimension with the advent of digital technology. Digital databases such as Canmore have immense potential for kinds of research which was simply impossible in the days of solely print publication, but problems of labelling and indexing constitute a barrier to realising that potential in full. The historic accumulation of data in these resources means there are inconsistencies in spelling and hyphenation in the database which can effect search results (e.g. searches on the variants 'ogam' and 'ogham', both of which are in scholarly and popular use, bring up different results). Technological improvements in 'fuzzy' searches may alleviate this, but some tidying up of data is necessary (see Canmore upgrade: Case Study 30; Reflections on Terminologies: Case Study 28). In some cases the terminology used is not that of the scholarly community, or brings up perverse results (e.g. 'gravestone'). This is a non-trivial problem which nonetheless must be addressed before Canmore can be used effectively to search for categories of carved stones. Scholars typically devise or refine terms in an ad hoc and intuitive way to meet a particular research need, and terms may not be conceived within an overarching hierarchy or with an awareness of established principles of classification and terminology. A collaborative approach—involving scholars, IT specialists, archivists and non-academic end-users—is necessary to establish agreed principles, in consultation with cognate bodies elsewhere. A project of this nature is likely to be a substantial undertaking but it has the potential, not only to greatly enhance the kinds of research that is currently undertaken but also to facilitate new types of research based on data searches which were hitherto unthinkable. These terminological issues are not only a concern for Canmore users, but also beset those who search for carved stones related items and new discoveries in the indexes of publications such as the annual Discovery and Excavation in Scotland, and Proceedings of the Society of Antiquaries of Scotland.
Two final points: in addition to English language taxonomies and terminologies generated by modern scholars, due attention should be directed to nuances of meaning enshrined in traditional terms relating to carved stones in Gaelic (e.g. clach, leac, coirthe) and Scots (e.g. lair-stane, bore stone). Often these are preserved in place names which, in addition to yielding insights into past understandings of carved stones, can sometimes provide important evidence of the prior existence of lost stones or the original/previous location of moved stones.
A separate issue from the labelling of recognised types of monuments is the naming and numbering of individual stones, and the lack of recognised conventions for doing so. With the possible exception of Roman stones, which are typically referred to in the scholarly literature using one of two authoritative numbering schemes enshrined in major corpora, comprehensive numbering schemes are not much used and there is a lack of consistency in naming stones (variously from the name of the field, nearest farm, nearest settlement, or the parish). Stones which have been moved may be referred to by the name of their findspot and/or the name of their current location. The extent to which traditional names for stones are respected and used varies (e.g. the Pictish cross-slab referred to as, alternately 'Drumdurno'/'Chapel of Garioch'/'The Maiden Stone'). In some cases scholars have coined memorable names for crosses, e.g. Constantine's Cross' for the Dupplin cross, but this can cause problems if the invented nature of such names is lost sight of, e.g. the invented 'St Oran's Cross' in contrast to the traditional names for the crosses of saints John, Martin and Matthew.
Where there are multiple stones from a single site, their numbering may follow one of several logics (e.g. order of discovery/publication, distance from core of site, size/degree of completeness, order of museum accession) or be effectively arbitrary. Lost stones may or may not be included in numbering sequences, and problems also arise when separate numbers are assigned to fragments of what is, in reality a single monument. This lack of care and consistency in numbering and labelling schemes and the perpetuation of competing systems can lead to confusion regarding the identity and numbers of stones at sites, resulting in unwarranted omissions or duplications. The British Academy Corpus of Anglo-Saxon Sculpture usefully laid out principles for the consistent naming and numbering of sculpture and these have been followed by the Welsh corpus (i.e. parish name, site name, number by order of discovery; also separately numbered within county). If applied consistently, such a system could be usefully adopted for Scotland's early medieval sculpture and, perhaps with modifications, for other periods also. Canmore is in the process of moving from an entirely site-based system (in which multiple carved stones at a single site were registered collectively under that site) to one in which individual early medieval carved stones are each assigned their own entry (see Canmore upgrade: Case Study 30; Canmore upgrade example: Case Study 4). This will, in effect, give each early medieval carved stone a unique reference number which, although too cumbersome for ready labelling, will doubtless prove useful in keeping track of the 2000+ stones in this category. More recent categories of carved stone are perhaps too numerous to be individually catalogued in this way, though it may be practical for some sub-categories.