I have a 1.9A dataset that could, in principle, be used to a molecular replacement solution. Phaser finds a solution which yields reasonable map to model agreement, but R factors do not fall down from values greater than 40%. It looked like a P21 (systematic absences are clear), so there was not much options on symmetry. I reprocessed the dataset in P1 and ran phenix.xtriage, and these lines called my attention: --------------------------------------------- Analysing possible twin law : -h,-k,l --------------------------------------------- Results of the H-test on a-centric data: (Only 50.0% of the strongest twin pairs were used) mean |H| : 0.053 (0.50: untwinned; 0.0: 50% twinned) mean H^2 : 0.005 (0.33: untwinned; 0.0: 50% twinned) Estimation of twin fraction via mean |H|: 0.447 Estimation of twin fraction via cum. dist. of H: 0.449 --------------------------------------------- Analysing possible twin law : -h,-h+k,-l --------------------------------------------- Results of the H-test on a-centric data: (Only 50.0% of the strongest twin pairs were used) mean |H| : 0.093 (0.50: untwinned; 0.0: 50% twinned) mean H^2 : 0.015 (0.33: untwinned; 0.0: 50% twinned) Estimation of twin fraction via mean |H|: 0.407 Estimation of twin fraction via cum. dist. of H: 0.407 --------------------------------------------- Analysing possible twin law : h,h-k,-l --------------------------------------------- Results of the H-test on a-centric data: (Only 50.0% of the strongest twin pairs were used) mean |H| : 0.085 (0.50: untwinned; 0.0: 50% twinned) mean H^2 : 0.013 (0.33: untwinned; 0.0: 50% twinned) Estimation of twin fraction via mean |H|: 0.415 Estimation of twin fraction via cum. dist. of H: 0.417 ========= I have never dealt with twinned datasets before. Is there something I could try to do to obtain useful data from this? Lucas P.S.: Unfortunately, growing more crystals is not an option. Alertas do Yahoo! Mail em seu celular. Saiba mais em http://br.mobile.yahoo.com/mailalertas/