This paper deals with an index of skewness (the letter coefficient of skewness, l.c.s.), proposed in a previous study (Brizzi, 2000), which is based on a particular set of sample quantiles called letter values, introduced by Tukey (1977). Some analytical properties of this index are described, and the sample distribution of this index is simulated under symmetric (normal and uniform) and positively skewed (exponential, Gamma, Log-normal and Skew-normal) distribution models. The power of a test based on the l.c.s. value is then compared with the 'classic' test focused on Pearson third-moment skewness index.