Subject Access to Art Images
The Language of Images
It Begins with the Cataloguer
The Image User and the Search for Images
Annoted List of Tools
Selected Bibliography
Illustration Credits
Printer Friendly PDFs

Introduction to Art Image Access

The Image User and the Search for Images

Christine L. Sundt
Curator and Professor, Visual Resources, Architecture and Allied Arts Library
University of Oregon, Eugene

An image seeker is a motivated user for whom needing a picture of something is the ultimate motivation. Time, or its absence, often dictates the urgency of the need. Cost is another factor: whether the image is worth the price, whether time is more precious than money. Some users know exactly what they want, while others are simply shopping for a good fit between an idea and its representation. How they communicate their needs, what obstacles they encounter in their quest, and how to help them avoid common pitfalls are some of the key points I address in this essay.

People look for images today as they always have, but they also look for them in new and different ways. This "then" and "now" dichotomy can be restated as traditional or manual access versus online or digital access. Traditionally, an image seeker—with or without a specific image in mind—began by browsing through books and magazines, using resources within reach with or without a specific image in mind. If the search through the materials at hand proved unsuccessful, the investigation would widen: with the help of a skilled reference professional, the user could find catalogues, published indexes, or vertical files where pictures were stored for the convenience of the information professional, and sometimes the user. Access systems for these files were largely idiosyncratic—conceived and constructed by whoever was in charge of the materials—because classification systems for pictures were, and for the most part are still, lacking uniformity and conformity to any standard. With the help of a resourceful information professional gifted with a photographic memory, homegrown finding aids, and a share of good luck, the searcher's needs could usually be fulfilled.

Today's digital environment offers new possibilities—and new challenges—for the image seeker under the guise of technology that seems to empower the end-user. By using the World Wide Web search engines or connecting to the many online sites specializing in art, architecture, and cultural heritage resources, users can browse through images without going to a library or an image archive, or asking anyone for assistance. Remote access has its benefits, especially when a search can be undertaken at any time of the day or night, but ultimately the success of the search depends on the skills, knowledge, and luck of the seeker, and on how well the resources being searched have been constructed and indexed. Success can be achieved through this kind of "unmediated" research, but the chances of finding the best fit quickly and efficiently are often small. The results of some searches produce too many choices, or they may be incomplete and confusing.

With images playing a much greater role in our everyday lives than ever before, the user, even the experienced scholar, has to deal with many obstacles in the quest for an elusive image. The need for better avenues to image resources is still an unfulfilled dream for many. Over the past few decades, new tools have been created to assist with specialized terminology and complex subject descriptions, as described elsewhere in this book. These tools can take us a long way toward the goal of removing the language-based roadblocks, but only if they are implemented consistently by those who build information resources and utilized to their full potential by the user. The challenges for librarians, archivists, cataloguers, and developers of new tools that can assist users in accessing images are obvious:
  • To create interfaces that accommodate and guide end-users through either or both simple and complex queries for both known and unknown images
  • To build "knowledge bridges"—that is, to fill in the knowledge gaps between the user, the image, and the textual data used to describe the image
  • To recognize the complexities that are often inherent in the "document-in-hand"—in most cases the image itself —in developing access points to that "document"

Anatomy of the Image User

We know something about users through studies, but regrettably we still lack enough information to know everything about their needs. We should more closely scrutinize our user logs to discover what our users are looking for and the words and phrases they are using in formulating their searches. From a study by Linda H. Armitage and Peter G. B. Enser, we know that users' needs have been neglected as an area of serious inquiry. We also know that there are noticeable similarities in how people formulate queries even across a range of image disciplines, and we are told that we can better serve the user by embedding analytical "schemas" within the information interface.1

Another study, published as Object, Image, Inquiry: The Art Historian at Work, solicited opinions from eighteen scholars "to represent a broad sampling of art historians active in research."2 In the chapter entitled "The Process of Art-Historical Inquiry," regarding the relation between original works of art and reproductions, we learn that art historians are savvy image users who understand the vast differences that often exist between the object and its reproduction:

The manipulative nature of reproductions and the relative merits and deficiencies of various media demand wariness. Nevertheless the practical difficulties of traveling to and comparing original works have produced ingenious methods of using reproductions of all kinds. As works of art and as records of conservation history, older photographs have special value. Collections of reproductions, whether institutional or personal, constitute vital resources for the art historian. The comprehensiveness of the collection, the inclusion of less well-known works, the scholarly acumen used in cataloging, and the difficulty of obtaining photographs were common preoccupations. Another was the absence of context that the photographic reproduction imposes on the work of art.3

The art historian's method of finding data is more a gathering process than anything else. Scholars collect data by "plowing through heaps of stuff" just "to find one particular piece of information." To them the process has its rewards: "you find out a lot of other things that you would never know you should know about."4 The process of discovery, the looking and selecting, could sometimes prove to be more valuable than immediately retrieving a direct "hit."5

In general, scholars are well aware of the limitations of reproductions. In Object, Image, Inquiry, one said, "The work of art has a kind of object-hood and physical presence which is very different from any [reproductive] image; even if the slide or the transparency were perfect, it's third-best."6

Using Words to Look for Images

Users tend to approach an image search by specifying layers of information. However, they are generally unaware that they are actually setting up hierarchical relationships, and they often fail to understand why some methods work better than others in constructing a query.
  • "I'm looking for a picture of a group"
  • "I'd like it to be a family group"
  • "This family should be doing something that would be typical for a family, like sitting around a table with food in front of them, looking grateful for what they have to eat "

The hierarchy in this query demonstrates a thought process that proceeds from a general concept—the group—to the specific concept—a family sharing a meal together. One example of such a family group might be Vincent van Gogh's The Potato Eaters (1885), in which a peasant family from the Dutch town of Nuenen, wearing traditional costumes of the time and place, are seated around a table.7 The room in which they have gathered is dimly lit, illuminated only by the solitary oil lamp above them. Their humble meal consists of potatoes.

Keywords can help users formulate queries like those listed above for submission to an online search engine, but their effectiveness varies widely. The particular searcher's cultural background, education, and even verbal skills condition the choice of keywords used in an online search, not to mention his or her native language. Most objects and concepts can be described using multiple terms or phrases in many languages or dialects. If an end-user happens to use a keyword different from the one in the information system being searched, he or she may miss items that are actually there. As the authors of the other essays in this publication have shown, controlled vocabularies and thesauri can be enormously powerful tools for bridging these kinds of verbal gaps.

Even with accompanying texts or captions, images can still be difficult to locate. Captions that do not follow a good "tombstone" template (that is, artist, artist nationality and life dates, title, medium, creation date, repository) may describe the meaning or interpretation of the image but not state the facts about the artist or creator of the object represented in the image. Many images appear online without any reference to what they are or what they represent. Again, the other authors of this book have stressed how important indexing and cataloguing are in enabling end-users to find images by using keywords or textual strings.

Simplicity and common sense should be the guiding principles in finding solutions to assisting end-users:
  • Know the needs of your users
  • Employ simple yet effective user interfaces
  • Err on the side of more rather than fewer access points
  • Know what the tools employed to assist the user are designed to do, as well as their limitations
  • Remember that what works in one situation may not be equally effective in another

The Quest

Consider for a moment where we see images. Much of our world looks different today, since television and computer monitors changed from monochrome to high color; since text phrases became icons; since books with few, mostly black-and-white images turned into richly illustrated color publications; and since our home printers began offering us the option of output in either black-and-white or color. These changes seem to have come about rather quickly, and yet some things have not changed at all. Consider now the challenge of finding a specific image or any image, an image of quality, an image that can be acquired for use without major limitations.

The search for the right image is still one of the more challenging exercises that users face, and perhaps more so since image use is at an all-time high. Thanks to technology, images are everywhere and seemingly available to everyone; image sites have sprung up all over the World Wide Web. The image may be ubiquitous, but the way we look for images today remains very much the challenge it was in the past. What might be even more alarming is the fact that finding the right image can also be more complicated now because, while we have much to choose from, our access resources and discovery skills are still quite primitive.

The Approach

In many instances, an image seeker has a preconceived idea of the desired image—a specific object or artwork, a place, a mood, a concept, a color, or a vision that is only a glimmer in the mind's eye. Finding a representation that fulfills the notion of the desire often takes time and skill. The searcher's success often depends on luck and perseverance.

"I'm looking for an image of ." is the usual starting point for this type of query. If the searcher can use well-crafted descriptive phrases and explain the nature of the needed image, then the results may be easier to obtain. Choosing the right words to describe the image and offering a context for the image or its use may provide useful starting points. The family seated around a table in van Gogh's painting might have been difficult to pinpoint unless some of the accessories were named. By adding keywords such as "potatoes," "lamp," "family," "eating," "meal," and "peasants," we can provide access via the main elements of The Potato Eaters.

We saw in Patricia Harpring's essay that an image of Herakles can be associated with a variety of themes: "Greek hero," "king," "strength," "fortitude," "perseverance," "labors," "Argos," "Thebes." In other words, this image can have many uses. Had the image of Herakles been indexed with this array of search terms, the user would have no trouble finding it as long as one of the indexing terms was used in the search. It would not be necessary to remember the name Herakles—or its variant, Hercules—to bring results; the search would produce the image from any of the other terms associated with it.

Now suppose that a user wants an image of "labor." While the word seems specific enough, what matters to the user is how labor is represented. Looking for an image of "labor" is, therefore, not just a matter of locating a record that happens to include the term. The user may not want an image of Herakles' famed labors, but rather something having to do with the Labor Movement in post–World War I America—two very different subjects. This difference in the meaning and use of the term "labor" reflects the distinctions between the identification of the image and the interpretation of the subject, steps beyond the literal meaning of the term in its context, as noted in the other three essays in this volume.

If the user has a specific image of "labor" in mind, then the task can be more challenging when trying to find an equivalent to an elusive memory imprint from times past. So often we find that our memories are flawed; that what we remembered as one color was in reality another when we finally recovered the elusive object. How many times have you discovered that the blue book you were looking for actually had a red cover? Similarly, we may be remembering a detail of a whole, an image that apparently had sufficient power to stand on its own but does not warrant a unique identifier as a proper title. Conversely, without a fixed image in mind, the user is more open to choices. The right image emerges on the basis of "I'll know it when I see it." This could be the case with the user searching for an image to represent the Labor Movement. When none of the images retrieved shows the exact historical moment, the user finds that choosing another that conveys the spirit of the movement satisfies the need. Of course, if the end-user types in the keyword "labor" and the resource being searched uses the British spelling "labour" (or vice versa), relevant items could again be missed, unless a thesaurus that includes alternate spellings (as the Art & Architecture Thesaurus [AAT] does) is used or alternate spellings are included as indexing terms attached to the particular item.

Another option is to look for an image based on a title or written description that seems to include all the right elements for a perfect representation. How surprising when the words do not fit the picture—when the words actually have little connection to their meaning but are used to represent an abstract concept or to convey personal meaning. Robert Motherwell's series Elegy to the Spanish Republic exemplifies the distance between words and image in that the abstract columns and bulbous forms rendered in stark contrast to each other—in many paintings from the series, black against a white field—are not taken from the facts of the Spanish Civil War but rather are a particular artist's reaction to the idea of human loss, resistance, and an ongoing struggle.8 We can only know that the title does not illustrate a specific event in history by knowing the artist and his oeuvre, by knowing that Motherwell would not be a likely source for a factual illustration of a historical event.

So where does this leave the user and us in trying to locate an image? Obviously there are common pitfalls that snare a seeker of a particular image. How can these best be avoided?

Access Points

An image is more than a subject or a title. As the Motherwell example illustrates, it may be important to know something about the artist or the designer of the object depicted in the image. It may also be helpful to know when it was created, what the medium of the work is, who owns it, where it is located or displayed, the circumstances surrounding its making, how large it is, and whether it was ever altered. As Colum Hourihane points out in his essay in this volume, the two criteria employed by most online searchers appear to be subject matter and creator. These starting points are codified among several standard description tools used by museums and libraries, as summarized in the metadata standards crosswalk mentioned by Patricia Harpring.9 One of these metadata standards, Categories for the Description of Works of Art (CDWA), has been used in examples elsewhere in this volume to show how subjects depicted in works of art are deciphered and described. But Subject Matter is only one element of a CDWA description. Its other core categories include Creator, Creation Date, Materials and Techniques, Measurements, and Current Location (see the CDWA record for a Panathenaic amphora on page 29).

Not every work can be described to the extent outlined in CDWA (nor would this necessarily even be desirable were it practical), but any data that follow a standard description format, where controlled vocabulary or terminology can be applied, are ultimately more accessible than data that do not adhere to any standards or vocabulary control. The value of the data are still largely conditioned by the skill of the indexer and the rules governing the data entry process, however. As Hourihane has argued, not all cataloguing and indexing are equal; nor are all data records complete or even correct.

Nonetheless, a descriptive record that includes only the title or description of a work may not be sufficient for providing access to its image. More promising would be, for example, a record that includes a date or time span that puts the title into a historical context and then adds information about its medium to help differentiate between works that are two- and three-dimensional.

Size or scale can also be valuable in helping to differentiate works bearing the same title, by the same artist, of the same date and medium, where one is likely a smaller model from which the larger finished work was created. Another useful element is the current location or ownership information about the work. With this information, the user is equipped with names and places: where to go for more information about the object, where it can be viewed, or where and how to obtain a reproduction of it.

A Case in Point

Searching for an image of Lot and His Daughters, a biblical subject (Genesis 19:30–38), one finds that the seventeenth-century Italian artist Orazio Gentileschi painted not one but at least five finished versions of this theme. One of these, dated to 1622, is in the collection of the J. Paul Getty Museum (pl. 7).

The Getty's picture shows a sleeping man, Lot, dressed in a blue garment, between two women, his daughters. The sisters' gazes and gestures lead us toward the right side of the canvas, to an event happening in the distance. They sit in front of a dark rock, presumably the cave where they and Lot had taken refuge, with their backs mostly to the viewer, one more in profile than the other. To the left of the daughter in profile are metal vessels, one a silver flask on its side, open and apparently empty, and the other a golden cup. These "props," in combination with the figures in this setting, are keys to the iconography of the painting.

Both women wear garments, but the one on the viewer's right and farther to the back is shown with bare skin, where her dress has fallen off her shoulders. The background includes ominous clouds and a bright glow above the distant hills. The glow obviously refers to the fire consuming the cities of Sodom and Gomorrah from where they had fled, but not before Lot's wife was punished for looking back as they were leaving. Lot's daughters, believing that they were the last human beings to remain on earth, have made their father drunk prior to sleeping with him—to save the human race. This portrayal of incest was popular in Gentileschi's time because of the artistic and erotic liberties it offered artists and their patrons. In the Getty's version, the daughter on the viewer's left wears a red garment over a white shirt; her sister is clad in a golden yellow chemise.

A second work by Gentileschi bearing the same title is in the Gemäldegalerie in Berlin (see pl. 8). The main difference between this work and the Getty's picture is that the colors of the daughters' garments are reversed: the daughter in profile wears a golden yellow dress and the one on the right is dressed in red.

Another version, now in the National Gallery of Canada/Musée des beaux-arts du Canada in Ottawa (see pl. 8), was purchased in 1965 from the Spencer Churchill collection in London. The colors of the daughters' garments are similar to the Getty's version, but the metal objects in the foreground are missing. Also, the background sky and landscape seem less ominous, the burning city is missing, and the overall contrast of light and dark (chiaroscuro) is understated compared to the other examples. The cave behind the family group is larger and rounder than in the other paintings, and the foreground rock cluster shows smoother edges and larger masses as well. Finally, where the foliage growing among the rocks in the other examples is alive and bushy, in this version it is a just a branch, devoid of leaves.

A fourth version is part of the Fundacion Colección Thyssen-Bornemisza in Madrid (see pl. 8). In some sources, the painting is said to be located in Castagnola or Lugano, not in Madrid, but still owned by Thyssen-Bornemisza. The daughters' garments follow the color pattern in the Getty's version. Data recorded in a sampling of contemporary literature about these four versions of Lot and His Daughters are presented in the table on page 76.10

There is a fifth version under this title, also oil on canvas. The largest in the group, measuring 226 3 282.5 cm, it is held by the Museo de Bellas Artes, Bilbao (fig. 20).11 This last work, signed by the artist and dated by scholars to 1628, shows a different arrangement of the daughters and their father, as well as the setting, inside rather than outside the cave, so it is chiefly related to the other four versions by its title and creator.

The Flemish artist Lucas Vorsterman (1595–1675) made an engraving of the Bilbao version, the plate presumably produced under Orazio Gentileschi's supervision in London sometime in the 1630s. The print, a reverse of the painted image, measures 332 3 430 mm and is in the collection of the British Museum in London.12 Vorsterman's engraving is just one of many copies and reproductions made by artists after the Bilbao and other versions of Gentileschi's composition.

Why So Many Versions?

It may be comforting to know that if one needs an image of Lot and His Daughters, there are at least five paintings by Orazio Gentileschi and many copies after his work to choose from. Perhaps any of the versions will suffice, but it may be beneficial to have options. It also may be a source of confusion, given what we have discovered about this composition and its scholarly sources.

Why would an artist paint more than one picture of the same subject? It may be that the image was popular and several of the artist's patrons wanted copies. Or it may be that the patron was dissatisfied with one version and wanted something changed, which resulted in another painting, or several more, before the patron was happy with the commission. Still another reason might be that some of the works are by followers or students of the artist—"practice pieces" from a later date. In this example, all five paintings are believed to be by Orazio Gentileschi himself.

Only a careful study of all the facts known about a work of art will bring an informed answer. This type of study relies largely on a combination of the verity of information that accompanies an image of the work and a careful analysis of the work itself—looking at the condition of the paint, how the paint was applied to the surface, how the support was constructed, whether the work shows signs of alterations (sections or pieces added or removed), and, perhaps of greatest value, the history of the work's ownership, its "pedigree" or provenance. CDWA facilitates the collection of all these facts in a consistent and orderly manner so that someone studying the work will find rich, interrelated data associated with the object. But even scholars are sometimes mistaken by evidence in hand.


The artist sometimes assigns a title to a work of art or architecture, but a curator or scholar who has carefully studied the work often assigns it after the fact. How titles are assigned and what they mean in providing access to a work of art are issues discussed elsewhere in this book and at considerable length in CDWA. We know from the Gentileschi composition that the figures, setting, and props contributed to our identification of the iconography as belonging to the story of Lot from the Old Testament. In looking for an image of an object or artwork, we must take into account the fact that titles can vary, especially if there is a question about the subject matter.

In the Gentileschi examples, all the works bear the same title, even though one painting shows a different arrangement of elements and personages. The same applies to Motherwell's series, where all the images are named Elegy to the Spanish Republic but each shows a different composition. A good example of a single work with drastically different titles is Rembrandt's famous painting, The Night Watch (1642; Amsterdam, Rijksmuseum).13 We now know that a better title for the painting is The Militia Company of Captain Frans Banning Cocq. Still another title might include the names of the various people depicted in the scene.

The title The Night Watch, in fact, was given to this work because of the somewhat dark varnish that once covered the painting's surface; the scene looked to be happening under the cover of night. Once the painting was cleaned and studied again, more facts emerged, requiring that a new title be assigned. Since the change occurred relatively recently (during the 1940s), most of the literature refers to the painting as The Night Watch. Thus, it is perfectly plausible that an image in a publication dating from the 1930s would be identified only with the old title. Without knowing more about the work and its alternate title, a person looking for an image may miss a valuable cache of reproductions and information, disconnected by time from the more recent research about the same work of art.

The guidelines given in CDWA accommodate and even encourage the inclusion of alternate titles and names. The indexer should seek out as many title variations as possible when describing a work of art and use tools that bring these variables together. Structured vocabularies and thesauri such as the ones discussed at length in the other essays in this volume were specifically designed to address the problem of variable terms and names for objects, media, creators, and places, and to assist the cataloguer in creating relationships among the variants. Structured vocabularies and thesauri, such as the Library of Congress's Name Authorities and the AAT, are useful for some aspects of subject description (usually those not dealing with strictly narrative or iconographic content), but a tool such as ICONCLASS is necessary for describing the narrative or iconographic meaning of works of art. As demonstrated by Harpring and Hourihane in their essays, a system like ICONCLASS—or a carefully constructed local authority file of subjects—can be used to create hierarchical relationships among iconographic themes or narrative episodes. It can also be used to make connections among images with similar compositions, where figures are grouped in a like manner or where accessories, furnishings, and props appearing in the work are identical. For example, even though the metal vessels are missing from the National Gallery version of Lot and His Daughters, the composition would be linked to the others in the series because of other compositional similarities. Properly analyzed and described in this way, two seemingly unrelated works can be reunited, or one work based closely on another can be recognized.

Measurements and Dimensions

At least three versions of Gentileschi's Lot and His Daughters are closely related not only via their titles and compositions but also by their dimensions. In addition to variations among versions, researchers should note whenever a single work in the group has been described with different measurements. The variation may be due to conversion between inches and metric measurements or because one person measuring the work took the numbers from inside the frame (so-called sight measurements), while another measured the canvas with the frame removed. One person may measure rounding off numbers, while another is more precise.

Measurements can be misleading in other ways as well. Some prints (engravings, etchings, and aquatints, for example) are measured to record the size of the plate from which the impression or relief was taken, while others record the measurements of the full sheet of paper carrying the print. Sculpture measurements can vary as well. Height can be determined by measuring a statue or object with or without its base or pedestal. In the case of ancient sculpture, heads often are reattached to torsos at a later time. The measurements of a statue may include later additions, and sometimes even restored parts. Measurements should be used with caution in critical comparisons or when trying to prove that two works of art are the same or different.

Dates and Dating

The four analogous Gentileschi images have been assigned similar dates, between 1621 and 1624, but the artist himself dated none of them. Many dates, or perhaps none at all, can be associated with a work of art or architecture. Attempting to find an image based on the date of an object might be difficult unless that date is highly significant to the work of art. Sometimes even dates that appear on the work are suspect; they could have been added by a later hand or included to refer to an event preceding the creation of the work. Scholars who have spent considerable time reconstructing an artist's oeuvre, making distinctions between the artistic styles of one expressive period and another, often assign dates. Perhaps some of the more remarkable dates can be found in the dating of Greek pottery, especially works of Attic origin. For these works, artists are assigned names, for example, the Meleager Painter,14 and their styles are placed within a chronological construct that defines the birth, adolescence, maturity, and ultimate decline of this art form. Few works within this construct are firmly dated or even signed, but the literature is rich in seemingly precise dates. Some of the more difficult areas to define according to date are Etruscan art and the artifacts of native tribal cultures. Nevertheless, dates or date ranges are often given simply as a way of differentiating one style or period from another.

Location, Location!

Many works of art have the potential to be mobile. Even a fresco that was originally part of a narrative or decorative cycle and affixed to a wall can find its way into a museum and be displayed as an independent work.15 Entire buildings can be housed within a museum; for example, a Maori tribal house is now in the Field Museum of Natural History in Chicago, thousands of miles away from its original location in New Zealand.

The popularity of eBay and art auctions is not a new phenomenon. Works of art and other artifacts move now, as in the past, from one owner to another at the drop of a hammer (or click of a computer mouse). The literature about the Gentileschi paintings demonstrates that the works have had many owners. Some owners housed their collections in different cities (Lugano, Castagnola, and Madrid); and scholars, even after careful analysis, can disagree on the exact pattern of ownership for some works. Three of the four versions of Lot and His Daughters were acquired by their current owners since the 1960s, so even some fairly recent literature includes references to former owners. A case in point is the Ottawa version, which was housed in the Spencer Churchill collection in London prior to being purchased in 1965 by the National Gallery of Canada. Often it is possible to link a work to an earlier reference by comparing its physical features (measurements, surface blemishes, and so on), as well as closely inspecting any available reproductions. However, an image can be deceptive and untrustworthy, too, as we saw in the art historian's warning about reproductions quoted at the beginning of this essay (see p. 69).

A work shown only in black-and-white or monochrome reproductions can be misleading, since subtleties of color are lost. In a catalogue raisonné of Gentileschi's work, the canvas now owned by the Getty was misidentified as the Thyssen-Bornemisza version because when two black-and-white reproductions were compared, they appeared to be the same work (fig. 21).16 The differences that could be perceived were thought to be the result of restoration—one showing the painting before restoration, the other after. In reality, the photograph shows yet another version, the Getty's, not included in the catalogue raisonné.17

The analysis was further hampered by the fact that the subtle differences between the colors of the daughters' garments—reddish on the left and golden yellow on the right in the Getty, Thyssen-Bornemisza, and National Gallery versions, but the reverse in the Gemäldegalerie painting—are not immediately visible in the monochrome reproductions. Users unaware of these color variations might not realize that this important distinction exists, or they might not realize that reproductions can be inaccurate due to the photographic process and the limitations of early black-and-white film in rendering color. 18

Color reproductions can be equally misleading. The same work of art shown in two color illustrations can look entirely different if the overall color balance is off. In sum, judgments based on the examination of reproductions rather than the study of the actual work can lead to false conclusions. Image seekers should be warned about drawing assumptions based on reproductions. Indeed, even the best may only be "third rate."

Again with regard to location, architectural elements have moved from one place to another throughout history. The famous Elgin Marbles, now in the British Museum in London, were removed from their original location on the Parthenon in Athens in the early nineteenth century. Before the invention of photography in the 1830s, pictorial records in the form of paintings, drawings, and prints provided evidence regarding the location and condition of works of art and architecture. For these non-photographic types of reproductions, artistic license and the skill of the artist who created a particular image played an important role in whether the rendition was ultimately accurate.

Another phenomenon is the changing of data about the location itself. National borders change due to political events, and countries take new names to reflect a new regime or newfound independence. Tools such as the Getty Thesaurus of Geographic Names enable us to link former names with a modern one (for example, Lisbon was called Ulixbone under Moorish rule, and Felicitas Julia under the Romans; the name Persia was officially changed to Iran in 1935, but it also refers to a region of what is now Southern Iran, known as Parsa, inhabited by Indo-European people around 1000 b.c.e.) and to reference geographic entities that no longer exist in the modern political world (for example, Etruria, Flanders, Holy Roman Empire, Phoenicia). Cities are also subject to remodeling, renaming, and annexation. Streets often have more than one name in use and perhaps several more buried in earlier directories. Buildings referenced by a street address must be studied in their historical and political contexts. Buildings, too, change over time. Consider the Louvre in Paris, with its various incarnations from fortress to palace to museum, and the many architects who contributed to its forms.

The Sum of Many Parts

Given the obstacles described above, locating a specific image can involve a considerable amount of work, even for an image seeker with considerable knowledge. Even when the object in question has a title, that title may not be an accurate reflection of content, as in the Motherwell example. A title may point to many works that are similar yet different enough to make selection of one difficult, as in the versions of Gentileschi's Lot and His Daughters. The date assigned to a work may be misleading, and the artwork or object may have been moved several times. The dimensions can vary from one source to another, seeming to suggest that the work must be a different one when in fact it is the same. An image without any accompanying descriptive data is virtually useless, however. Finally, the quality of the image—how accurately it is represented by its illustration—is an important factor in deciding whether to use a particular image.

The researcher or image seeker must know how to judge and balance the facts associated with a picture. Vocabulary tools and classification systems such as the ones discussed in the preceding essays provide valuable assistance in sorting through questions having to do with names, terms, and iconography. How well these tools are used in creating descriptive records becomes the deciding factor in the end. The more the cataloguer or indexer can do to facilitate access —through standards, common tools, and shared strategies—the easier it is for the searcher to find what is needed, be it one image or many. A picture may be worth a thousand words, but one hopes that the words themselves have value for the image seeker as well.