The content model for figures is designed to provide support for single or multi-part figures in two ways.
For a simple display in multiple parts, where a single graphic can be used to display the entire figure, there is no need to break the image into pieces.
For a larger, or more complex image, that will not properly display in a single graphic online, the figure should be broken into parts that are inserted as individual graphics within the wrapper figure. For example, an image that had parts a-h may be (arbitrarily) broken into three graphics representing three parts ("a-c", "d and e" and "f-h").
Keeping them all within the same figure element allows them to share the caption, and also provides for stacking the images vertically on the screen instead of displaying one huge image (the
<figure> element supports multiple
<graphic> elements). Each graphic is given a unique identifier that is the same as the figure identifier, but with an additional sequence letter suffix:
<figure id="F003">
<caption>Caption For All 3 Graphics</caption>
<graphic entityref="F0003a.gif" id="F0003a" ... />
<graphic entityref="F0003b.gif" id="F0003b" ... />
<graphic entityref="F0003c.gif" id="F0003c" ... />
</figure>
These identifiers can be directly referenced from the text:
... seen in figures (3a-c) and (3f-h) ...
... seen in figures <figref linkend="F0003a">3a-c</figref> and
<figref linkend="F0003c">3f-h</figref> ...