HTML table advanced features and accessibility

You can give your table a caption by putting it inside a <caption> element and nesting that inside the <table> element. You should put it just below the opening <table> tag.

<table>
  <caption>
    Dinosaurs in the Jurassic period
  </caption>

  …
</table>

As you can infer from the brief example above, the caption is meant to contain a description of the table contents. This is useful for all readers wishing to get a quick idea of whether the table is useful to them as they scan the page, but particularly for blind users. Rather than have a screen reader read out the contents of many cells just to find out what the table is about, the user can rely on a caption and then decide whether or not to read the table in greater detail.

A caption is placed directly beneath the <table> tag.

Let's try this out, revisiting an example we first met in the previous article.

1. Open up your language teacher's school timetable from the end of HTML Table Basics, or make a local copy of our timetable-fixed.html file.
2. Add a suitable caption for the table.
3. Save your code and open it in a browser to see what it looks like.

As your tables get a bit more complex in structure, it is useful to give them more structural definition. One clear way to do this is by using <thead>, <tbody>, and <tfoot>, which allow you to mark up a header, body, and footer section for the table.

These elements don't make the table any more accessible to screen reader users, and don't result in any visual enhancement on their own. They are however very useful for styling and layout — acting as useful hooks for adding CSS to your table. To give you some interesting examples, in the case of a long table you could make the table header and footer repeat on every printed page, and you could make the table body display on a single page and have the contents available by scrolling up and down.

To use them, they should be included in the following order:

• The <thead> element must wrap the part of the table that is the header — this is usually the first row containing the column headings, but this is not necessarily always the case. If you are using <col>/<colgroup> elements, the table header should come just below those.
• The <tbody> element needs to wrap the main part of the table content that isn't the table header or footer.
• The <tfoot> element needs to wrap the part of the table that is the footer — this might be a final row with items in the previous rows summed, for example.

Let's put these new elements into action.

1. First of all, make a local copy of spending-record.html and minimal-table.css in a new folder.
2. Try opening it in a browser — You'll see that it looks OK, but it could stand to be improved. The "SUM" row that contains a summation of the spent amounts seems to be in the wrong place, and there are some details missing from the code.
3. Put the obvious headers row inside a <thead> element, the "SUM" row inside a <tfoot> element, and the rest of the content inside a <tbody> element.
4. Save and refresh, and you'll see that adding the <tfoot> element has caused the "SUM" row to go down to the bottom of the table.
5. Next, add a colspan attribute to make the "SUM" cell span across the first four columns, so the actual number appears at the bottom of the "Cost" column.
6. Let's add some simple extra styling to the table, to give you an idea of how useful these elements are for applying CSS. Inside the head of your HTML document, you'll see an empty <style> element. Inside this element, add the following lines of CSS code:
   css

   tbody {
     font-size: 95%;
     font-style: italic;
   }
   
   tfoot {
     font-weight: bold;
   }
   

7. Save and refresh, and have a look at the result. If the <tbody> and <tfoot> elements weren't in place, you'd have to write much more complicated selectors/rules to apply the same styling.

Your finished table should look something like the following:

It is possible to nest a table inside another one, as long as you include the complete structure, including the <table> element. This is generally not really advised, as it makes the markup more confusing and less accessible to screen reader users, and in many cases you might as well just insert extra cells/rows/columns into the existing table. It is however sometimes necessary, for example if you want to import content easily from other sources.

The following markup shows a simple nested table:

<table id="table1">
  <tr>
    <th>title1</th>
    <th>title2</th>
    <th>title3</th>
  </tr>
  <tr>
    <td id="nested">
      <table id="table2">
        <tr>
          <td>cell1</td>
          <td>cell2</td>
          <td>cell3</td>
        </tr>
      </table>
    </td>
    <td>cell2</td>
    <td>cell3</td>
  </tr>
  <tr>
    <td>cell4</td>
    <td>cell5</td>
    <td>cell6</td>
  </tr>
</table>

The output of which looks something like this:

table {
  border-collapse: collapse;
}
td,
th {
  border: 1px solid black;
  padding: 10px 20px;
}

Let's recap briefly on how we use data tables. A table can be a handy tool, for giving us quick access to data and allowing us to look up different values. For example, it takes only a short glance at the table below to find out how many rings were sold in Gent during August 2016. To understand its information we make visual associations between the data in this table and its column and/or row headers.

But what if you cannot make those visual associations? How then can you read a table like the above? Visually impaired people often use a screen reader that reads out information on web pages to them. This is no problem when you're reading plain text but interpreting a table can be quite a challenge for a blind person. Nevertheless, with the proper markup we can replace visual associations by programmatic ones.

This section of the article provides further techniques for making tables as accessible as possible.

Screen readers will identify all headers and use them to make programmatic associations between those headers and the cells they relate to. The combination of column and row headers will identify and interpret the data in each cell so that screen reader users can interpret the table similarly to how a sighted user does.

We already covered headers in our previous article — see Adding headers with <th> elements.

A new topic for this article is the scope attribute, which can be added to the <th> element to tell screen readers exactly what cells the header is a header for — is it a header for the row it is in, or the column, for example? Looking back to our spending record example from earlier on, you could unambiguously define the column headers as column headers like this:

<thead>
  <tr>
    <th scope="col">Purchase</th>
    <th scope="col">Location</th>
    <th scope="col">Date</th>
    <th scope="col">Evaluation</th>
    <th scope="col">Cost (€)</th>
  </tr>
</thead>

And each row could have a header defined like this (if we added row headers as well as column headers):

<tr>
  <th scope="row">Haircut</th>
  <td>Hairdresser</td>
  <td>12/09</td>
  <td>Great idea</td>
  <td>30</td>
</tr>

Screen readers will recognize markup structured like this, and allow their users to read out the entire column or row at once, for example.

scope has two more possible values — colgroup and rowgroup. These are used for headings that sit over the top of multiple columns or rows. If you look back at the "Items Sold August 2016" table at the start of this section of the article, you'll see that the "Clothes" cell sits above the "Trousers", "Skirts", and "Dresses" cells. All of these cells should be marked up as headers (<th>), but "Clothes" is a heading that sits over the top and defines the other three subheadings. "Clothes" therefore should get an attribute of scope="colgroup", whereas the others would get an attribute of scope="col":

<thead>
  <tr>
    <th colspan="3" scope="colgroup">Clothes</th>
  </tr>
  <tr>
    <th scope="col">Trousers</th>
    <th scope="col">Skirts</th>
    <th scope="col">Dresses</th>
  </tr>
</thead>

The same applies to headers for multiple grouped rows. Take another look at the "Items Sold August 2016" table, this time focusing on the rows with the "Amsterdam" and "Utrecht" headers (<th>). You'll notice that the "The Netherlands" header, also marked up as a <th> element, spans both rows, being the heading for the other two subheadings. Therefore, scope="rowgroup" should be specified on this header cell to help screen readers create the correct associations:

<tr>
  <th rowspan="2" scope="rowgroup">The Netherlands</th>
  <th scope="row">Amsterdam</th>
  <td>89</td>
  <td>34</td>
  <td>69</td>
</tr>
<tr>
  <th scope="row">Utrecht</th>
  <td>80</td>
  <td>12</td>
  <td>43</td>
</tr>

An alternative to using the scope attribute is to use id and headers attributes to create associations between headers and cells.

The headers attribute takes a list of unordered, space-separated strings, each corresponding to the unique id of the <th> elements that provide headings for either a data cell (<td> element) or another header cell (<th> element).

This gives your HTML table an explicit definition of the position of each cell in the table, defined by the header(s) for each column and row it is part of, kind of like a spreadsheet. For it to work well, the table really needs both column and row headers.

Returning to our "Items Sold August 2016" example, we can use the id and headers attributes as follows:

1. Add a unique id to each <th> element in the table.
2. Add a headers attribute to each <th> element that acts as a subheading, i.e., has a header element above it. The value is the id of the heading that sits over the top and defines the subheadings, which is "clothes" for the column headers and "belgium" for the row header in our example.
3. Add a headers attribute to each <td> element and add the ids of the associated <th> element(s) in form of a space-separated list. You can proceed as you would in a spreadsheet: Find the data cell and search for the corresponding headings for the row and column. The order of the specified ids doesn't matter, but you should be consistent to keep it organized.

<thead>
  <tr>
    <th id="clothes" colspan="3">Clothes</th>
  </tr>
  <tr>
    <th id="trousers" headers="clothes">Trousers</th>
    <th id="skirts" headers="clothes">Skirts</th>
    <th id="dresses" headers="clothes">Dresses</th>
  </tr>
</thead>
<tbody>
  <tr>
    <th id="belgium" rowspan="3">Belgium</th>
    <th id="antwerp" headers="belgium">Antwerp</th>
    <td headers="antwerp belgium clothes trousers">56</td>
    <td headers="antwerp belgium clothes skirts">22</td>
    <td headers="antwerp belgium clothes dresses">43</td>
  </tr>
</tbody>

1. For this final exercise, we'd like you to first make local copies of items-sold.html and minimal-table.css, in a new directory.
2. Now try adding in the appropriate scope attributes to make this table more accessible.
3. Finally, try making another copy of the starter files, and this time make the table more accessible by creating precise and explicit associations using id and headers attributes.

There are a few other things you could learn about tables in HTML, but this is all you need to know for now. Next, you can test yourself with our HTML tables assessment. Have fun!

If you are already learning CSS and have done well on the assessment, you can move on and learn about styling HTML tables — see Styling tables.

If you want to get started with learning CSS, check out the CSS Learning Area!

• Previous
• Overview: HTML tables
• Next