Figures
Encoding of figures and illustrations using figure; handling of text within figures; discussion of the WWP’s changes to the content model of figure
Graphic features of the page which are representational (for instance, emblems, heraldry, portraits, illustrations, and the like) should be encoded using figure. In general, we feel it is unnecessary to encode non-representational ornamentation (such as page borders, section markers, decorative ruled lines, etc.) in this way, since figure provides more encoding overhead than is necessary to capture such features. See ornaments for our recommendation on encoding non-representational ornamentation.
Within figure, we strongly recommend using figDesc to capture a brief description of the figure; this serves the same purpose as the alt attribute in HTML, and is important to provide accessibility to the visually handicapped, as well as to provide keyword searching on images.
If you are capturing a digital image of every page, then there may be no need to provide a link to a specific image file representing the figure. However, if you are only capturing digital images of things like illustrations, the figure element should include a link to the corresponding image. In unmodified TEI, this is done using the entity attribute on figure, which points to an entity which in turn points to the image file (see example 1). A simpler approach is now in common use, in which a url attribute is added to the figure element, allowing you to point directly to a specific file location (see example 2). This modification is included in the DTD extensions that accompany this Guide.
Illustrations are often accompanied by text which is actually part of the figure itself, not simply part of the running prose that surrounds it. This text may take the form of a caption or heading, or it may be a short passage of prose or poetry either at the top or bottom of the image. Text may also appear within the image’s graphic content: for instance, a speech bubble, sign, banner, or inscription on something depicted in the image (such as a building, a gravestone, a piece of clothing, a book). The DTD extensions that accompany this Guide allow for slightly more flexibility in encoding this text, to provide for all of the different cases that appear in early modern printed books, as follows:
- Any headings associated with the figure should be encoded within head before the figDesc element.
- Any text appearing within the graphic itself should be transcribed in full within a text element following the figDesc. It may also be useful to briefly quote this text within the figDesc element, to show where the text appears; this is particularly true if you do not provide a digital image, so that users must rely on your description to visualize the figure. If there is more than one separate piece of text within the figure, each piece should be transcribed in text within a separate p or lg element. Pieces which are visually separate but logically continuous (e.g. Hail to on one side and the chief on the other) should be transcribed as a single unit within text.
- Any captions or epigraphs associated directly with the figure, or printed at the top or bottom of the figure (possibly even inside its frame) should be encoded within a p or lg element preceding or following the figDesc element. The question of whether a piece of text should be encoded in this way, or within the text element, will sometimes be a judgment call. The text within text should be only that which is truly mingled with the image content of the figure. It may also sometimes be difficult to determine whether text immediately preceding or following the figure should be associated with the figure at all: that is, whether it should be encoded within the figure element in one of the ways described above, or should be encoded as part of the ordinary stream of the text. This will necessarily be a judgment call. Text which is associated directly with the figure—not just referring to it, but logically associated with it—should be encoded within figure. The criterion for whether something is associated directly with a figure is whether it would naturally be included with the figure if the figure were excerpted from the text. (Compare examples 1 and 2.)
Examples
Example 1:
<figure>
<figDesc>An engraved portrait of Elizabeth I carrying
an olive branch, with a map of the Americas in the background,
and the words <quote>Honi soit qui mal y pense</quote>
embroidered onto her sleeve.</figdesc>
<text><body><p>Honi soit qui mal y pense</p></body></text>
<p>Our Gratious Queene, in the 22 yeere of her rayne</p>
</figure>
Note that within <gi>text</gi>, only <gi>body</gi> and <gi>p</gi>
are required (no <gi>div</gi> necessary).
Example 2:
<div><p>A noted Entomologist, who says he combed the nether
regions of his Garden in search of the Wooly Centipede,
at last concluded that this fugitive Insect spawns only
in the Winter, with a Grub in the following form:
<figure><head>The Grub of the Wooly Centipede</head>
<figDesc>A woodcut of a centipede larva curled up
on a leaf</figDesc>
<p rend="slant(italic)">Centipedula Lanifera</p>
</figure>
Its legs do not develop until the summer, when it assumes
its familiar shape. </p></div>