search for: chondrit

Hi, I am trying to create what are called "spider diagrams" in the geochemical literature using R. A spider diagram is basically a plot of the atomic number versus the concentration on a log scale. Lines are drawn from each atomic number for each sample. Right now, my data frame looks like: SAMPLE SITE V3 LA.NASC LAATNUM CE.NASC CEATMNUM 1 1A:001

Dear R users, I've written the following functions to implement Silverman's modality test ("Using kernel density estimates to investigate multimodality", J.R. Stat. Soc. B, 43, (1981), 97-99.), and I tested them on the Chondrite data set (Good & Gaskins, J. Amer. Stat. Ass., 75, (1980), 42-56). Values for the critical window width seem OK, which is not the case for the significance levels. If someone could give me a hint about what is wrong... Or perhaps someone has already done a real implementation of this test?...