Randall Munroe at XKCD did a color names survey a few years ago, and published a list of about a thousand colour names whose RGB values (averaged across his readers’ monitors) could be fairly reliably estimated.
I have finally got around to turning them into an R package. It’s only on GitHub so far. The functions are
xcolors(max_rank=-1): List the top (most commonly given) max_rank color names; analogous to colors()
name2color(name, exact = TRUE, hex_only = TRUE, n = -1): Find colours exactly or partly matching a given name, reporting just the hex color string or the full data, optionally only using the top n colours.
nearest_named(color, hex_only = FALSE, max_rank = -1, Lab = TRUE): given a matrix of RGB or a vector of color hex strings, report the nearest colours with names, optionally restricting to the top n colors, and using either Lab or R’s RGB colour space to define ‘nearest’.
For example, the nearest named colors to the standard palette are
>  "black" "fire engine red" "vibrant green" "primary blue" >  "cyan" "bright magenta" "bright yellow" "silver"
Restricted to the top 100 colours they are
>  "black" "red" "green" "violet" "cyan" >  "bright purple" "yellow" "grey"
The picture at the top of this post shows the ColorBrewer “Dark2″ palette, with the nearest named colors in both RGB and Lab space.