339In a previous report to this journal (1963) we made conclusions on the processes of speciation and production of endemics among the Darwin finches in the Galapagos Archipelago. The conclusions were inferred from multiple-regression analysis wherein four independent variables of the insular environment were quantified and tested for respective abilities to predict species data. The analysis was an attempt by linear regression to discover (a) the environmental determinants of insular variation in the species numbers (Y ,) and (b) the determinants of such variation for numbers of endemic subspecies (Y2 ) . Insular area (X,) and number of land plant species (X2 ) -presumably indexing ecologic and floristic diversity-were found lacking in power to "push" or "move" the dependent variables (the Y's). Isolation, measured by distance between islands (X3) and by distance from a given island to Indefatigable Island near the center of the archipelago (X,), was the major predictor of number of species and endemics for islands, with X 3 being more important than X, in terms of contribution to Y variances (~y2). For both problem a and b, however, a component (~dy2) of the variance remained unaccounted for by regression of the V's on the X's, and this was attributed to error inherent to the analysis, to factors not considered, or to both.1 Publication supported by the U. S. Public Health Service.To decrease the values of~dy2, we have (i) shifted from use of desk calculator to digital computer analysis; (ii) included in the analysis new environmental factors (Xr., X.) not previously considered; and (iii) utilized new models or estimating equations involving transformations of the arithmetic values of the primary measurements of the V's or X's to their logarithmic counterparts in nonlinear or joint linear-nonlinear variation. Below we summarize results of such a comparative multiple-regression analysis, and, following this, a statement is made of the bearing of the new information on our previous conclusions (Hamilton and Rubinoff, 1963). Data for the V's and the four X's are from our previous report, and the new factors (Xr., X.) are from a comparable study of plant species abundance in the archipelago (Hamilton et al., 1963). The first is multiple linear regression and the last three are examples of multiple nonlinear analyses. With inclusion of new factors, the number of