A histogram is a graphical display for a single quantitative variable. The range of the quantitative variable is divided into intervals called bins which are plotted on the x-axis. Then either the frequency (number) or relative frequency (percentage) of data values falling within each bin is represented by the height of the bar sitting over the bin.
A histogram is a good way to assess the shape of a distribution:
- Modes — Is the distribution unimodal, bimodal, or multimodal?
- Symmetry — Is the distribution symmetric or skewed?
- Outliers — Are there any outliers?
A very basic definition and some simple examples are given here. The Wikipedia page on histograms gives a definition and also includes a fairly technical discussion of choosing bin widths. (Mostly we let software make that choice for us.) This site has a java applet that allows you to play with the bin sizes for a number of interesting data sets. Finally, this site explains the pitfalls of confusing a histogram with a bar chart.
The Wolfram MathWorld site has a poor example of a histogram. (There should be no spaces between the bars of a histogram.)