Variant: Modeling concentration data with imprecise nuisance lag

Dear Nmusers, We have a problem that is similar to that mentioned in "Modeling concentration data with imprecise sampling time". In a tumor model, we measure 3 concentrations simultaneously with resolution in the seconds-range and high density. The dosing compartment 1 is measured with a different and less accurate method. For technical reasons the start of the time scale cannot accurately be adjusted, it has a nuisance jitter which we must compensate for to avoid trouble with very steep initial slopes. I tried something like: IF (CMT .NE.1) ALAG1 = EXP(MU_18+ETA(18)) IF (CMT .EQ.1) ALAG1 = EXP(MU_19+ETA(19)) but this does not work for MU/ETA 18 (it diverges), because dosing is in CMT 1. Janet Wade recommended TSCALE, but if I understand the documentation, this is a scaling (=multiplication of t), not a shift (=addition of t). Is there something like TSHIFT? Any ideas would be appreciated. Dieter Menne U Hospital of Zurich