On Tue, 10 Nov 2015, Sven Schreiber wrote: This is tricky. In the first case the object being assigned to a list is a named series from the dataset, which is OK. In the second case it's the series product of a calculation, and in general that's not acceptable; for example you cannot say list L = x/3 where x is an existing series. Now a human observer can see that in your case above the "calculation" happens to be transparent and just passes through one or other of two named series. I'll have to think if there's a way for gretl to recognize that without breaking other things. Consider this superficially similar-looking example: list lulu = (z==1) ? LRY : LRM where z is a series. Here the calculation is _not_ transparent and there is a genuine type mismatch: trying to add an anonymous, on-the-fly series to a list. Allin