This research is a classic case of "armchair ecology". Ecologists have been amassing data on all sorts of topics for decades, and we are now increasingly seeing ecological papers that are reporting on quantitative overviews of the literature. Some of these papers go as far as a formal "meta-analysis" of the literature in the manner propounded in medical science. My paper is less rigorous than that, but it produced some interesting generalizations. For example, it seems that habitats richer in native species are likely to have more, not fewer, invasive species. Islands have most plant invaders for a given area of ground, and mainland parks and reserves the least. One would hope this would allow us to draw conclusions about relative susceptibility to invasion of the different environments (their "invasibility"). However, conclusions about process based on pattern are unsound. The conceptual part of my paper argued that the number of invasive species established in a region is a product of the number of species introduced and the rate at which they survive in the new habitat. Thus, my early drafts of the paper in which I attempted to use the data on patterns of invasive species richness to construe invasibility (which is inversely proportional to the rate of species survival) were flawed. We cannot tell whether differences between habitats in invasive species counts have resulted from differences in invasibility or differences in the number of species introduced. This might seem pretty obvious now, and perhaps would have been to others, but I toiled on and off over 5 years analyzing and reanalyzing the data before I realized I was asking questions the data could not answer! This account would be incomplete if I did not thank the editorial staff at the journal, Ecology, for allowing me to persist with this paper through a couple of major revisions. They and the referees contributed substantially to its final form.