Figure
also called figcaption (community), figure caption (community), captioned image (community)
An image, table or code block bound to its caption so the two travel together and can be referred to from the running text.
A figure is a piece of content that the running text points at rather than contains. The test is whether it can be moved: if the paragraph still reads when the picture is lifted to the top of the page or the facing column, it is a figure, and it takes its caption with it wherever it goes. That binding is the whole component. Everything else, the border, the number, the placement, follows from the fact that these two pieces are one unit and the surrounding prose is not.
It is not only for photographs. A table, a code listing, a diagram, a chart, or a
short quotation all qualify, which is why the HTML element is figure rather than
something image shaped, and why the caption sits in a figcaption that may come
before or after the content it labels. Wiring it up properly buys something real: a
figure with a figcaption takes that caption as its accessible name, so a screen
reader announces the group by its label instead of dropping the reader into an
unattributed image in the middle of an article.
The caption is not alt text and writing one as if it were the other is the common mistake. Alt text describes the picture to someone who cannot see it, and it is redundant the moment the same words are printed underneath. A caption explains what the picture means to everyone: what is being shown, when, why it is here. So the alt text says a metal staff gauge stands in shallow water beside a stone quay, and the caption says the reading was taken at low water on the day the survey began. If the image is pure decoration it is a decorative image and it wants neither. And a caption in this sense is unrelated to captions, which are the timed text of a recording.
Number a figure only if the text refers to the number, and if it does, use the reference rather than the position: “the gauge in Figure 3” survives a layout change that “the picture below” does not. Captions are read far more often than body copy, because scanners look at pictures first, so write them as sentences with the point in them and never leave a filename or a bare source line standing in for one. Where the figure is the reason the reader stopped, the caption is the only text you can count on being read.
Which word?
| If you want | say |
|---|---|
| a visual needs a caption that belongs to it | figure |
| the picture adds nothing a reader would miss | decorative image |
| the image is data and one line will not do | long description |
| a picture is content rather than decoration | image |