Agreed w/most here - I like my comparables to generally reflect the subject's characteristics. That said, the acreage bookends will obviously 'grow' as the subject site gets larger. If I have a 2 acre site, I might go 1-5 for comps (5 being over 100% difference from the subject). If I have a 20 acre site, however, I'd probably only go 10-30 (which is only 50% of the subject's size each way). Regardless, I solve for all other variables (isolating the site size adjustment), and perform sensitivity analysis to determine the appropriate adjustment factor. Of course, if there are other influences on site value (external influences), I'd try to adjust for those before isolating the site adjustment itself.