The Hand Tools and Implements Collection

Interim Progress Report :: 1 May 1996

Note: This is an HTML document and may be retrieved on-line from URL


  1. Introduction
    1. Objectives of the Project
    2. Objectives of the Catalogue
    3. Implementation to Date
    4. File List
  2. Issues and Decisions
    1. Associations
      1. External
      2. Internal
    2. Access
      1. Verbal
      2. Visual
      3. Conceptual
      4. Analogical
  3. Unresolved Problems


Objectives of the Project

In conformance with the objectives of ILS 626, my goals in creating the Tools and Implements (TANDI) catalogue project have been:

Objectives of the Catalogue

The Tool and Implements catalogue, as it has been designed and to some extent implemented, is the concrete expression of a selected few of the tool-catalogue desiderata within a specific web implementation. I have been chiefly concerned:

Note that another statement of the information described by these two lists, which I have referred to throughout as "access" and "association" information, can be found in the pages attached to the information page, viz., the access page and the associations and contexts page .

Implementation to Date

Given the ambitious nature of these objectives, it should come as no surprise that their implementation has lagged somewhat behind. To date, the following items and features are in place.

  1. Bibliography
    A substantial subject bibliography of upwards of a hundred items has been assembled and examined (and several dozen otherwise unavailable books purchased!), devoted to the history of tools generally as well as to individual categories of tools (drafting tools, planes, surgical instruments, etc.), and consisting of contemporary trade literature, contemporary treatises and manuals, modern historical treatments, exhibition catalogs, museum catalogs, and similar works.

    In addition, several works (e.g. LCSH, the Art and Architecture Thesaurus, the Nomenclature museum cataloging system) have been examined with a view to establishing vocabulary control for both generic titles (uniform titles) and "subjects" (trade, craft, function names, etc.). None has proven adequate in itself, but may become so when used in combination with the "native" systems employed by contemporary literatures, especially trade literature.

  2. Information
    Hundreds of individual items of contextual information have been identified in the works listed in the bibliography.

  3. Images
    Perhaps a hundred photographs have been taken of tools in my possession, as well of a select group of tools borrowed for the purpose from the historical medical artifacts collection held by the University of Michigan.

    From these, as well as from many of the contextual pieces that happened to be illustrated, perhaps 150 scanned images have been taken for web use, edited in TIFF format, and converted into GIFs and (occasionally) JPEG-format images.

  4. Sample Entries
    Some fifty items have been assigned titles and alternative titles, and have received descriptive cataloguing, including the noting of model numbers, etc. On some, but not all, the author-name authority control work has been done, but much more needs to be done with respect to the names of both individual designers and corporate manufacturers. For web presentation, the information has been presented tabularly, separate sections being assigned to title, designer, alternative titles, manufacture date and responsibility, date of manufacture (often very much guesswork), physical description, dimensions, and materials, marks such as maker's marks and labels, model number, patent number, provenance information for the local copy, and variations or alterations shown by the local specimen.

  5. Subject headings Many of the entries have also been supplied with "subject" headings of the various kinds suggested by the access points listed above, provisionally including:

  6. "Other Information" system
    An expandable system of "other information" links has been begun, allowing a single link to be attached to each entry in each of several categories, each of which is supplied with an appropriate (as my imagination allowed) icon.

Though the system is in place, not all of its features have been implemented, and none has been implemented with complete consistency, among the fifty sample records. Most items have been supplied with photos and reference material, several with "mates" information, three with MARC records, three with "spoor" links, half a dozen with historical links, and so on.

It has not yet been finally decided how to handle two situations that this system is likely to encounter: two or more items linking to the same piece of information, and one item having two more links to information in the same category (one or more of which may itself be shared with some other record). The first situation is particularly likely to arise in the case of the "mates" and "spoor" links: all late nineteenth-century scalpels and bistouries in the collection, for example, might suitably be linked to an 1889 chart of available scalpels and bistouries. My provisional solution is to link all the relevant entries to a single page, and to insert return paths to all the linked entries on that page as the links are added, despite the rather ad hoc nature of this solution. A fully automated system, beyond my technical expertise, one that (e.g.) generated html on the fly from database information, would render this question moot: the relevant page would be generated automatically each time the link was chosen. The second situation has been dealt with, also provisionally, by the web standard method of adding a level to the path: if all the historical information related to an item cannot be accommodated by a single page, it is placed on several pages, all linked through a single page that identifes the multiple materials available. Technically speaking, high-resolution images present a similar problem of levels. Most of the images in this demo version have been kept low-resolution and fairly small: in a couple of instances (the amputation kit and the carving kit), I have allowed myself the luxury of a high resolutio image linked through a low-resolution thumbnail on the "reference" or "photo" link page.

Partial Implementation

To date, the following items and features have been designed but are not yet fully implemented (if at all):

  1. Information Links
    Enough information links have been provided to demonstrate the potential of the system, but many more are possible and planned.

  2. Subject Headings
    Some MeSH subject headings have been given to the medical-instrument entries, but more are planned, as well as headings from the Art and Architecture Thesaurus and LCSH. The necessity of using multiple sources of headings is part of a larger, not yet entirely solved, problem of...

  3. Controlled Vocabulary
    There is no single existing controlled vocabulary that will cover the collection in its entire scope with adequate specificity, much less one that allows the use of both general and specific terms simultaneously (essential to successful searching). Terms are needed for each of my labelled subject categories (task, trade, object, manner), as well as for uniform titles (in the entries) and the (largely equivalent) generic names (for tree-based searching or browsing). My temporary solution has been to use thesaurus-based terms each in its own field (MeSH:...; AAT:...; LCSH:...; Nomenclature:... ), though only MeSH terms are yet in place; adapt the open-ended Nomenclature system (little different from inventing my own vocabulary) for uniform titles, with alterations and additions based largely on the most commonly used or most readily available reference works and trade catalogues; and develop a local vocabulary for the other fields based on all of the above.

    The development of even a small controlled vocabulary is a substantial undertaking and I have as yet only begun to try to reconcile the differences between my sources.

  4. Interlinking
    With hypertextual linking such a distinctive feature and powerful resource accorded by the web environment, the catalogue should take advantage of it, not only by linking information to items but by linking items together as they share characteristics. In particular it should be possible, having identified a relevant entry, to

  5. Visual indexing
    This is one of the most important features of the system, one of the most pleasant to work on, and so of course the one left till last to implement in any but a very skeletal form. As explained in the Access Page, many tools and implements are readily recognizable, identifiable, and searchable only through a visual rendition of some distinctive feature. Examples include cross-sections of otherwise identical draw knives and broad axes, joints of forceps, contours of plane irons, shapes of inset boxwood strips in wooden plane stocks, stamped makers' marks on many tools, etc. This a feature that should in time be implemented at several points in the system, so that someone doing a tree search, for example, and having reached the branch that leads to "planes, molding," would be offered a choice of a verbal list of plane titles, a list of plane pattern names ("Grecian Ogee," "Quirk Ogee," "Astragal," "Quirk Ogee with Bead," "Grecian Ovolo with Bead," "Reverse and Back Ogee with Square," etc. -- another problem of controlled vocabulary and precise identification), or a visual index of contours, the last in many circumstances by far the easiest to deal with. The same feature should be available as an interlink feature, so that someone finding a "pipe wrench" should be able to click on that, and be offered a choice of a verbal or a visual index of tools belonging to that category. My sole demonstration trial of visual indexing has been put in place as an interlink on the terms....

  6. Boolean and proximity fielded searches
    Since I lack the expertise to create a script smart enough to call up a real search, I have deferred creating search forms. In keeping with the fine particularization of data in the records, the system should be able to search in the way made standard by many "advanced" web-based searches, that is by allowing a Boolean combination of string or proximity searches each within a selected set of fields.

  7. Image-gallery browsing and searching
    Searching by thumbnail has had to wait till I gained more expertise with images (this was my first experience with a scanner...). In the fully implemented system, image browsing, like visual indexing, should be available at several points in the system. One should be able to browse the entire collection visually (though this is not very scalable, and is not very useful as a search technique). More usefully, one should be permitted a visual browse option in response to any search response. Searching for "Mfr=Stanley," for example, should produce a page reading something like the following:
                 34 Records matched this request.
                 Do you wish to (choose one)
                   [LINK] View a title list of these records? 
                   [LINK] Scroll through the full records?
                   [LINK] View thumbnail pictures of these items?
    Each image should link, of course, to the full record for the item.


    To date, the following objective is still in the planning stage, and has not yet found a suitable embodiment in a workable design:

    File List

    The structure of the demonstration site is laid out by means of the following linked files, available beginning at my SILS/SI index page, and following the link there. Some lower-level files are not shown in this schematic.

    Issues and Decisions

    Faced with such a bewildering array of desiderata as regards record type, access possibilities, search mechanisms, and flexible and expandable information linking, the chief issues and decisions have arisen from the need to accomplish all of these with a feasible system that could be implemented in the real world and designed, at least in outline, in one semester. One decision that I made at this nitty-gritty level, a defensible decision, I think, was to take inspiration from the Ohio medical artifacts cataloging project mentioned above. In doing the actual cataloging, I have followed the procedures developed by that project and promulgated in the Dittrick Museum's Cataloguing Manual (listed in the bibliography). One might view my project, at least potentially, as an attempt to adapt procedures developed for the AACR2/MARC cataloging of medical instruments on OCLC to wider use, applying it to a wider range of instruments and tools, but preserving its basic regard for bibliographic standards as a vehicle for access to artifact collections. As Linda J. Evans points out (Beyond the Book, 187ff), the bibliographic standards in general and MARC in particular are not particularly well suited to this use. I have not attempted to make MARC carry the burden of my additional subject categories (beyond those used by the Dittrick), but there is no reason that it should not do so, using the "local standards" options available in the subject fields, in addition to the allowable MeSH, AAT, and (?) Nomenclature terms.

    Some of the procedures borrowed from Ohio include:

    Beyond the procedural, important though that may be, lies the basic concept of the system. Only within an overall concept do individual parts make sense. The disparate desiderata and procedural distractions have made achieving this vision the most challenging portion of the assignment. As I conceive of the system in its interaction with users, the design must afford a clear (if not always an easy) route to any item searched for, a path inward from the world in which tools are contextually embedded to the appropriate records and the tools in the collection for which they are surrogates. It must also provide a way to re-establish context, once that route has been traversed and a tool or set of tools identified, including both information resident in the system, such as other relevant items, and information brought in from outside, such as often requires expert subject knowledge to identify on one's own.


    This latter category, that of associations, connections, and contexts, has ended up taking a preponderance of my time, to the detriment of the access function. The fundamental decision made in this respect, perhaps a troublesome decision, was to leave the catalogue scalable and expandable and to allow uneven treatment of the collection. This seemed to me to be in keeping with the actual management of museum collections, as well as of archival ones, in which items deemed of little interest receive scanty treatment, while items deemed significant receive much more extensive treatment. It is designed furthermore to be a dynamic catalogue: as additional documentation is created for special events (exhibitions, or even a user's research), the results can be integrated immediately into the catalog. In fact, the catalog is flexible and powerful enough that it is even possible to create an exhibition within the catalog itself and use it to generate a printed or on-line exhibition catalogue, should that be desired, simply by preselecting the items and the links to be included.

    External connections

    One problem with imagining such a system, however, has been that in order to demonstrate it at all, I have needed to accumulate a great mass of subject knowledge; to exploit its resources fully, anyone would. I have needed to cease to be simply a librarian but also a tool collector, historian of surgery, and mechanic (and this in one semester). Still, it is vital that a museum cataloguing system be open ended, allowing encyclopedic treatment without (impossibly) insisting on it. It is possible to build this catalogue into a reference book or even an amalgam of all existing reference books on technological history; it is up to curator or librarian to draw the line on the external linking of relevant material, whether on a collection-wide or (more reasonably) ad hoc basis.

    Internal Connections

    The ability to make internal links among the items in the collection has also proven to be both potentially powerful and potentially troublesome. On the positive side, the ability to attach "mates" or "historical" information, or to add subject headings at will to items deemed to be in some way related, (especially the former) permits expert knowledge such as might be gained by a curator or reference archivist to be actually incorporated within the system; allows someone giving a presentation, for example, to ascertain that toolA is a direct antecedent of toolB in the history of that technology, to say so in a lecture, and then to add the information to the catalog for the benefit of future users.

    On the negative side, of course, such freedom can be exhausting to both cataloguer and user, a growing bulk of half-useful references distracting, and a tangled web of internal links a fair representation of chaos (just imagine what would happen if items had to be renumbered or renamed for some reason).


    The inward route, on the other hand, though potentially even more complex, has two radically simplifying aspects: it must always end with an item actually in the collection, not in the informational world at large, and its underlying structure is verbal; in time, controlledly verbal. As it appears to the user and seeker, of course, this is not obvious: the inward route appears to be multiple and to offer alternatives between verbal, visual, conceptual and analogical paths.

    Verbal access

    Verbal access is afforded by fielded text searching with truncation, adjacency, and Boolean options. Generous use of headings and titles makes cold searches (i.e. searches without use of a thesaurus) unusually productive; most users will need choose no other path, since recall should be high. The field-limitation and other options permit precision to be high as well. The verbal option will be the most obvious choice, of course, when the pre-existing knowledge is also verbal: usually, names, whether of persons, places, companies, trades, or tools.

    Visual access

    Visual access is afforded through two means: image browsing and visual indexing of distinctive features. Both are necessary (though the latter perhaps more so), because the functional nature and often ingenious origin of tools and related artifacts introduces in them small but vital differences searchable only by feature indexing, even while their general appearance, representable in a photograph, manifests a memorable, even an aesthetic character, that can most readily be searched for by an image-based mechanism.

    Conceptual access

    Conceptual access, by which I here mean access via some sort of abstract consideration of function or means, though susceptible to verbal approaches, will be most directly served by the tree-based searching system that I have, unfortunately, had no success yet in implementing.

    Analogical access

    This brings us back to internal contextualizing, since by "analogical access" I mean that means of access that begins with the statement, 'I am looking for something that resembles this.' Implementation of interlinking, whether verbal or visual, is designed specifically to afford a capacity for analogical access.

    All of these processes of access depend on headings and labels that exist only in verbal form. A visual index of drawing knives depends for its association of those records on the title "drawing knife"; an image gallery that includes a machete and a pruning knife but not a scalpel depends on some such subject heading as "agricultural tools and equipment (to use the Nomenclature designation); one that includes a machete and a scalpel but not a winnow, on a designation like "edged tools." And the conceptual tree search is of course only a thesaurus hierarchically displayed and equipped with links. At the root of the catalogue, then, lies the power and the problem of controlled vocabulary and of naming practices generally. Museum cataloguers, as Evans points out (191), "must rely on their own traditions in formulating titles," a situation that has both resulted from and contributed to the lack of national artifact catalogs. In MARC terms, fields 655, 755, and 740, among the most important for artifact cataloguing, remain and (in the nature of things) must remain ill served by comprehensive thesauri. Even Nomenclature, the one thesaurus specifically aimed at the comprehensive naming of arifacts, is less a full list than a hierarchy, a naming method, and a preliminary list to which individual collections are invited to add local terms (Nomenclature, 16-7). My experience with Nomenclature suggests some of its deficiencies as well as its strength. It includes, for example, the surgical "trocar" --but only as a veterinary instrument; it does include, surprisingly, most of the specific plane types included among the sample entries -- plane, smoothing; plane, rabbet; plane, fillester -- but not the cooper's howel (perhaps regarded as the same as a compass plane) or cooper's leveler or sun plane (perhaps the same as a cooper's stoup plane); a scalpel but not a bistoury, a tenaculum but not an aneurism needle; a plumb bob ("unclassified tools and equipment") and a hog ringer ("ringer, animal-nose; animal husbandry tools and equipment") but not a "plierench" (though pliers may be subdivided by local terms).My own attempts to augment this list have depended heavily on trade literature: Tiemann, Truax, and Sharp & Smith, for example, for medical instruments; or the mammoth McMaster-Carr catalogue, alongside the Wards and Sears catalogues (which Nomenclature itself used), for tools and equipment generally. My decision has been to begin with Nomenclature and their techniques and hierarchy, at least for object names and hierarchical placement, and to expand from there, as they invite users to do.

    Classification by "subject" (a debatable use of the MARC subject fields, by the way), such as trade, task, or function, is another matter, better handled by the general subject thesauri (AAT, MeSH, LCSH). Used in combination, these not only produce a mish-mash but leave large gaps among the trades especially, and among the trades of earlier times even more especially. Again, I have started with the standard thesauri but resorted to secondary for additional terms: some trade literature divides by trade and function or medical specialty, and other sources such as exhibition catalogues, industrial surveys (the famous "Manufactures of Sheffield" in the 18th century), and even treatises on heraldry (which divides all of society in its occupations and roles) can provide surprisingly helpful.

    Unresolved Problems

    There is clearly much to be done, even aside from technical implementation, to bring this project into an acceptable intellectual state. The policy on vocabulary control, informal till now, must be made rigorous--must itself become subject to control. Policy on the use of MARC fields should be tightened up. What, exactly, counts as a subject or a title or a form or genre? And the many sections that are only in the planning stage need to be tested in practice. Though it should be apparent that this is very much a work in progress (and I am sorry to present it in such a state), I believe that the staggering breadth and complexity of task that I discovered that I had set myself, with its need to consider thesaurus creation, image managment, multiple means of access, encyclopedic linking, web technology, MARC formatting, and so on, all within a slowly developing concept of the desired end product--as well as the need to develop practical skills like file management, scanning and image management, and HTML along the way--more than account for its condition.

    More positively, the pieces are in place for a catalog that represents a substantial advance, especially as regards remote access and flexibility, on most of those in the field.

    Paul F. Schaffner :: 1 May 1996