Hi, I recently used columns of a matrix as inputs to nls(), which works just fine, but then I get, for example, (assuming >nls(stuff)->nl30 was the original command) >nl30$call nls(dr[,2]~apowr^(dr[,1])/(b*sqrt(dr[,1]))) Now, I can build a new matrix with same dimensions as dr and use that to run predict, as in >predict(nl30,list(dr=newdr)) , but I was wondering if there's a way to edit the contents of nl30 itself? I changed the contents of nl30$call just fine (using expression() on an edited version of the formula itself), but that does not change whatever items in nl30 are used by predict(). Again, I recognize that I'm much better off running nls() with more sensible independent and dependent variables to begin with. I just would like some help to learn how to read the "guts" of the class variable that nls() produces. (yes, I know how to do "nl30$m" and so on; those lists don't show me any of the variables or formulas I can get using formula(nl30) and other commands with methods for this type of class variable). Thanks Carl