The Smithsonian Institution
The Smithsonian Institution is an important national asset. A vast organization, it comprises 16 separate museums and 8 research centers with combined collections totaling over 240 million artifacts 150 million in the museums themselves, the rest in associated libraries and archives. Fewer than 2 percent of the collection is on display at any one time. The federal appropriation for the Smithsonian in fiscal year 2001 was $454.9 million [Bley 2001].
The developers of the Smithsonian's Web site faced a huge challenge. They had to create a way to access a dauntingly huge and diverse collection representing the digitized holdings of multiple component museums with different missions and different types of collections. At the same time, they had to create an aesthetically and intellectually rich experience for visitors of many different backgrounds and capabilities, including scholars and casual tourists from all over the world and people of all ages and educational backgrounds.
On our first visit to the Smithsonian site in July 2001, we were disappointed to encounter a number of significant accessibility barriers. As a national museum, the Smithsonian receives significant federal funding, as mentioned above, and we assumed that the Smithsonian Web site was required to meet Section 508 accessibility standards. Our assumption was wrong, however: the Smithsonian is not bound by Section 508 because it is actually a National Trust, not an agency of the federal government. Nonetheless, we were aware that the Smithsonian had made significant efforts with regard to accessibility (including the exhibit on the disability rights movement discussed in Chapter 3 as well as the "Guidelines for Accessible Exhibition Design" cited earlier in this chapter), and we assumed that the site design would take the needs of people with disabilities into account. But our hopes were dashed pretty much right off the bat.
The site has undergone a major revision since that initial visit, but in July 2001, the home page at http://www.si.edu required that Flash 5 be installed on the user's computer in order to access a con tinually changing list of links. Flash 5 objects are inaccessible to screen readers, refreshable Braille displays, and text-only browsers, yet the information in the Flash window was not available in any other way. The Smithsonian home page as we found it on that first visit also spawned a confusing pop-up window that presented an advertisement for a current exhibit at one of the Smithsonian museums. That pop-up window included a form where visitors could "register" for the Smithsonian site; we had the mistaken impression that we would have to register in order to use the Smithsonian site from this point on but the <input> elements contained within the <form> element had no <label> elements to identify them, and the form was inaccessible to people using screen readers, talking browsers, refreshable Braille displays, and text-only browsers. Finally, the page included numerous images (<img> elements) without the required alt attributes.
We found other serious problems during that first visit as we moved into the "interior" of the Smithsonian Web site. The online collections portion of the site, then available at http://digilib.si.edu/digilib/main.asp but taken down sometime around mid-October 2001, used a template including four frames with name attributes like body and navbody that had little meaning for the user. Links to information about the artifacts displayed on these pages were unintelligible, consisting only of database record numbers like the ones on the Amazon.com site discussed in Chapters 2 and 8. Onscreen text identified the artifacts, but it was difficult for someone using a screen reader to determine which label was associated with which object.
The Smithsonian Web site was redesigned in spring 2002; as a result, we cannot show you screen shots of the site as it appeared throughout 2001. The new home page at http://www.si.edu solves the problems we encountered on our first visit the Flash animation and the pop-up window with its unlabeled form have been banished, and all <img> elements have the required alt attributes. But other problems remain unsolved, and new ones have been introduced. A brief discussion of selected Smithsonian Web sites will show you what we mean.
The Smithsonian's HistoryWired Site
Launched in August 2001, the Smithsonian's site, HistoryWired: A Few of Our Favorite Things (Figure 7-1), offers a promising approach to the challenge of showcasing a vast collection in a way that takes account of visitors' interests as they change over time for example, in response to current events.
Figure 7-1. Screen shot of the introductory page of the HistoryWired section of the new Smithsonian Web site. Accessed June 10, 2002, at http://historywired.si.edu/index.html. Used with permission.
At any given time, HistoryWired offers 450 highlights from the Smithsonian's Museum of American History; the artifacts on display change as visitors express their preferences. To accomplish this, the HistoryWired site uses technology first developed by Smart Money magazine to help investors. Smart Money's programmers adapted the technology so that people visiting HistoryWired can affect its development by expressing their preferences. The visitor can rate any artifact; the higher the rating an artifact receives, the more space the category to which it belongs is given on the site. It's a great idea!
But, as it launches, the page immediately spawns a pop-up window just like the old Smithsonian home page. Anecdotal evidence suggests that most people find such pop-ups annoying. But pop-ups can be bewildering for users who are blind, for users who have cognitive impairments, and for many elderly users as well, and WCAG 1.0 Checkpoint 10.1 makes it a Priority 1 requirement to avoid spawning new windows without first notifying the user.
The use of the pop-up technique is especially problematic in the case of HistoryWired because the pop-up window contains instructions on how to use the site's interactive tools, including the Java "map" that provides a visual indication of the strength of visitors' interests. This map is inaccessible to people using screen readers, talking browsers, refreshable Braille displays, and text-only browsers, so perhaps it's fitting (if a bit ironic) that the instructions for using it are presented via a technique that represents an accessibility barrier. The instructions in the pop-up window do refer to a text-only alternative to this map but the very users who most need that alternative version will miss the instructions. Putting crucial instructions in a pop-up window is a poor design decision for many reasons, not least of which is the fact that pop-ups are usually associated with highly intrusive and unwelcome advertisements, and many people simply close them as soon as they recognize that a pop-up has appeared.
The text-only version of HistoryWired (available at http://historywired.si.edu/text.cfm) is used for a legitimate reason: to provide access to functionality that cannot be made accessible in any other way (WCAG 1.0 Checkpoint 11.4). The text-only version isn't really text-only, however. As shown in Figure 7-2, once you choose a topic from the text-only index page and proceed, the pages display images of highlighted artifacts.
Figure 7-2. Screen shot of a page from the "text-only" version of the HistoryWired site. Accessed June 10, 2002, at http://www.historywired.si.edu/object.cfm?ID=342. Used with permission.
We wanted to learn about the ENIAC computer, the machine that launched contemporary digital computing. We were distressed to discover that neither the image of the ENIAC nor the related image has ALT text. (JAWS reports the larger of these two images as "link graphic objects/342a." It reports the other image as "thumbs/342a," which might lead users to the mistaken belief that it is a thumbnail version of the larger image. In fact it is a small picture of ENIAC's principal designer, J. Presper Eckert.)
We might have liked to see more artifacts like this one, but we had trouble expressing our preferences. The page includes a <form> element where, as explained on the screen, visitors can "vote" on whether they'd like to see "More" or "Fewer" objects like this one. Visitors then use radio buttons to indicate how strong their preferences are on a scale from 1 to 10. However, there are two problems.
The radio buttons are not associated with the instructions visible on the screen; JAWS reports only the radio button's label and status.
JAWS reports only one option, "Fewer," and reports it multiple times as you press the up or down arrow key. As a result, it's impossible to know what you're voting for or what rating you're assigning to it.
Chapter 10 provides information about how to make forms accessible. We will find, as we explore some other museum sites, that inaccessible forms are but one of many barriers encountered by seekers of scientific and cultural information who have disabilities or who use assistive devices to browse the Internet.
The Smithsonian's National Air and Space Museum Site
The National Air and Space Museum site at http://www.nasm.si.edu/ includes a link (the first on the page) called "text v". This link goes to a so-called "text-only" version of the site at http://www.nasm.si.edu/NASMhome_t.html (which in turn has a link to something called "icon version"). The "text-only" version of the site includes a graphical logo that apparently has no ALT text: JAWS reports it as "graphic NASMicon_60.jpg (12271 bytes)." When we checked the source code, however, we found that there was an alt attribute for this image after all but instead of saying something like "National Air and Space Museum logo" or "Home," it repeated the name of the image file. (See Chapter 9 for information about writing effective ALT text and other text equivalents.)
The Smithsonian's Hirshhorn Museum Site
There is a fascinating section called Art Interactive on the Hirshhorn Museum Web site. Located in Washington, DC, the Hirshhorn is an outstanding museum devoted to twentieth-century art; it is also one of the Smithsonian museums. Visitors can use the Art Interactive section to explore different types of art and learn about the materials artists have used and how the works are created at least some visitors can do so. But a visitor with a visual impairment who wants to learn about assemblages and found art (such as collages or sculptures made from discarded objects in which the artist recognized interesting possibilities) may be frustrated.
Following the Assemblage/Found Object link takes us to a page that contains content about the subject we chose but that, according to JAWS, is confusingly mistitled "Animals in Art" (Figure 7-4). As we access the content, we realize that we have indeed come to the page we intended, but the title read by JAWS has nothing whatever to do with that content. The page is dominated by a large image of a work by a French artist, Jean Dubuffet (1901-1985). The work is called "The Soul of Morvan"; according to the text on the page, it features grape wood and vines as well as rope, tar, and metal. The text goes on to explain that the artist "used grapevines from the wine-growing region of Morvan, France, to evoke a weatherworn man laboring in a vineyard. By leaving the materials in a rough, earthy state, Dubuffet made the sculpture even more expressive." While this text is useful, even evocative for people who know what grapevines look or feel like, it would be difficult for someone who cannot see the image on the screen to understand how grape wood, vines, rope, tar, and metal come together to create the image of a man and even more difficult to understand what any of this has to do with the topic Animals in Art.
Figure 7-4. Screen shot of a page from the Smithsonian's Hirshhorn Museum Web site. Accessed June 10, 2002, at http://hirshhorn.si.edu/education/interactive/assemblage.html.
The Hirshhorn site also includes a page titled Accessibility (at http://hirshhorn.si.edu/visit/accessibility.html). The page describes the location of parking spaces, restrooms, and other facilities that meet the physical needs of people with disabilities. But there is no reference at all to the accessibility of the Web site itself, and there are several untagged graphical links on the page; there is also a link called, weirdly, Void (0), which actually appears on all pages that we visited on the Hirshhorn site (appropriately enough, the Void link does not seem to work). Like all the other pages we visited on Smithsonian Web sites, the Hirshhorn's Accessibility page has no way to bypass repeated navigation links, as WCAG 1.0 Checkpoint 13.6 recommends. Consequently, people using screen readers, talking browsers, and refreshable Braille displays must plow through all the navigation links before reaching the substance of the page.
World Museums on the Web
Museums around the world face the same challenge as the Smithsonian: they have to find ways to represent their collections and make them available to users from diverse backgrounds and with a wide range of knowledge and skills. Many museum Web sites use graphical links as an interface to various parts of their collections. In many cases, the image of a typical artifact from the collection serves as a link to information about the history and extent of the collection, as well as digitized highlights; in some cases (such as London's Tate Gallery), the entire collection is available online. This can be an effective strategy, but it can also present accessibility barriers. Although both WCAG Checkpoint 1.1 and Section 508 paragraph (a) require text alternatives for images and other non-text elements, this basic requirement is overlooked on many museum sites.
On the Collections page of the Musée du Louvre in Paris (http://www.louvre.fr/), for example, a picture of the Venus de Milo is used as a link to a page that provides an overview of the collection of Greek and Roman antiquities and links to more detailed historical information and images. Someone who uses a screen reader, talking browser, refreshable Braille display, or text-only display when visiting the Louvre site has no way of knowing that this option is available, however. When the screen reader encounters the images on the Collections page, it announces that we are reading a table with eight columns and six rows that contains links such as "ao/ao_f," "ae/ae_f," "ager/ager_f," "sculp/ scu_f," "oa/oa_f," "peint/peint_f," "ag/ag_f," "palais/pal_f," and "aaoa/ aaoa_f." Which of these would you follow to learn more about Venus if by some chance someone had informed you to use the graphic image of that famous statue to get to the antiquities? The correct choice would be "ager/ageréf," an abbreviation for "Les antiquités grecques, étrusques, et romaines" (Greek, Etruscan, and Roman antiquities) but how could you know? The words "Antiquités grecques, étrusques, et romaines" are right there on the screen but if you use a screen reader you'll never hear them, because they're actually part of the image that includes the Venus; from a technical standpoint, they're not text at all but are instead part of the graphic. (See Chapter 9 for more about the distinction between text and images of text.) It would be easy for the Louvre's Web team to solve this problem by adding an alt attribute for each <img> element in this graphical menu; in this case, the ALT text might read "Les antiquités grecques, étrusques, et romaines" exactly the words that appear on the screen.
We ran into similar problems on the home page of the new Guggenheim Museum in Bilbao, Spain. The link to the Bilbao site on the Guggenheim's home page at http://www.guggenheim.org opens in a new browser window without notifying the user (WCAG 1.0 Checkpoint 10.1 warns against this). The visually elegant Bilbao site contains only four graphical links, none of which has a text alternative as required by WCAG 1.0 Checkpoint 1.1: JAWS identifies them only as "link graphic idioma 1," "link graphic idioma 2," "link graphic idioma 3," and "link graphic idioma 4." 
 Accessed June 10, 2002, at http://www.guggenheim-bilbao.es/idioma.htm.
Back in the United States, the Houston Museum of Natural Science Web site (http://www.hmns.org) contains a mix of textual and graphical links. Sighted users see a menu of textual links on the left side of the screen; these lead to information about events and programs at the museum, ticket purchases, camps, museum membership, and so on. Sighted users also see a number of prominent graphical links. On this site, both types of links present accessibility problems. The link text displayed on the screen is tiny and it's not easy to enlarge because the font size is "hard-coded" to 8 points by the attached Cascading Style Sheet (CSS), which makes it impossible for people using some graphical browsers to increase or decrease font sizes to meet their needs (violating WCAG Checkpoint 3.4; see Chapter 15 for a discussion of style sheets). But people using screen readers, talking browsers, or refreshable Braille displays may not get that far anyway: before they have the opportunity to read the text links, they have to sit through a list of 14 graphical links attached to <img> elements. Those graphical links should be associated with ALT text that says exactly what sighted users see on the screen. The ALT text is missing, however, so the screen reader announces, "link graphic c3a," "link graphic c4a," "link graphic c5a," "link graphic c6a," and so forth, all the way to "link graphic c16a." In the absence of ALT text for the graphical links, users without visual access have no way of knowing whether the text links are redundant or not; these users are excluded from important navigation options available to people who can navigate visually.
The Web site for the Boston Museum of Fine Arts (http://www.mfa.org) begins with a "splash screen" that displays a single, untagged image before automatically launching the home page (a practice that violates WCAG Checkpoint 7.4). The home page (like other pages on the site) makes heavy use of image elements without alt attributes. Searching the collections database for images by Claude Monet turned up 42 records but the links to those records were unintelligible. Instead of ALT text, they used only the database record numbers ("link graphic fif=zoom/123-34.fpx&obj=iip,1," says JAWS, and then "link graphic fif=zoom/123-18.fpx&obj=iip,1," and so on) alternating with links to additional information about each work.  But each of those links is labeled "More," so again it's impossible for people using screen readers, talking browsers, refreshable Braille displays, or text-only displays to keep track of where the links go.
 Accessed June 10, 2002, at http://www.mfa.org/artemis/results.asp.
Other problems we encountered on the sites we visited included image map links with missing ALT text for their <area> elements (the Louvre, Paris; the Museum of Science and Industry, Chicago; the Museum of Modern Art, New York); text too small to be easily legible and difficult or impossible to resize (the Whitney Museum of American Art, New York; the Tate Gallery, London); multiple pages with the same title but different content (the Louvre, Paris; the Whitney Museum of American Art, New York); use of multiple frames with titles that do not assist orientation and navigation (the Louvre, Paris); and the absence of any way to skip over the many links that are repeated on each page of each site (all sites). Table 7-1 provides a quick overview of the museum sites we visited and the accessibility barriers we encountered.
Table 7-1 has three columns. Column 1 is the name of the museum we visited online. Column 2 is the URL. In Column 3 are some brief and by no means comprehensive comments about barriers that we encountered. (A linearized version of the table appears in Appendix C.)
Table 7-1. Accessibility Barriers on Selected Museum Web Sites
Chicago Museum of Science and Industry
Graphical links have no meaningful information, for example, "link graphic button." Text links have redundant titles, making them difficult to sort and therefore to use by users with screen readers.
Houston Museum of Natural Science
No alternatives are provided for graphic information including important navigation links.
Musée du Louvre
Frames names provide no orientation information. Graphic image links appear without meaningful alt text.
Museum of Fine Arts Boston
Splash screen with untagged image automatically redirects to front page with pop-up window containing untagged images. Online collection database search is usable but returns unintelligible links that are database records.
National Museum of Australia
ALT text could be more meaningful, and there is an inaccessible calendar of events, but by and large, a fairly usable site.
Mislabeled forms, use of inaccessible Java elements, pop-up windows, lack of ALT text on image maps, and misleading page titles.
The Guggenheim (Bilbao, Spain)
Pop-up windows are unannounced, graphic links with no ALT text.
The Menil Collection
Unlabeled graphic links, unannounced auto-refresh to a home page that contains links labeled Button 6, Button 8, Button 7, Button 3, Button 4 not in numerical order!
Whitney Museum of American Art
Site has insufficient user control options (font size is specified in style sheets) and inconsistent use of alt tags for graphics.
[*] All sites were accessed on June 8, 2002, except for the Smithsonian site, accessed February 16, 2002.
While it's clear that developers of some museum Web sites (the National Museum of Australia, for example) are taking the needs of people with disabilities into account, the strongest impression we took away from our whirlwind tour of museums on the Web was that museums have yet to embrace Web accessibility as an important part of their mission to preserve the artistic and cultural heritage of the communities they serve.