Pat, > > One proviso on alignment of different time frames is that you cannot > > currently align two time frames if the alignment would involve the use of > > DUT1 (the difference between UTC and UT1). In practice this means that > > the supported timescales are divided up into two groups > > > > 1) TAI, UTC, TT, TDB, TCG, TCB > > 2) UT1, LAST, LMSG, GMST > > > > and you can only align a time frame with another from the same group. > > The important distinction is that group (1) really are time scales whereas > group (2) are angles that describe Earth rotation. In fact Earth Rotation > Angle (and local ERA) could usefully be added to group (2). At the moment, I've not emphasised this distinction in order to provide a consistent interface and model for the TimeFrame class. Refering to UT1, etc, as a "timescale" also seems to be common practice in things like SUN/67 and Arnold Rots STC schema. > The sidereal times in group (2) ought to be IAU 2000 compliant. Are they? It's as per fortran slalib, so I guess that means no? If I'm interpreting things correctly, the equation of the equinoxes is IAU 1994 and the UT1-GMST relation is IAU 1982. Keeping up with changing IAU recommendations could be a problem. Presumably a supposedly general purpose thing like TimeFrame should ideally support a range of historical conventions as well as current conventions, since there will be lots of data out there which use the historical conventions? David